ℝ

Series ISSN: 1938-1743

ASHLOCK
Fast Start
Series Editor: Steven G. Krantz, Washington University in St. Louis

Fast Start Advanced Calculus

Advanced
Daniel Ashlock, University of Guelph

FAST START ADVANCED CALCULUS

This book continues the material in two early Fast Start calculus volumes to include multivariate
calculus, sequences and series, and a variety of additional applications. These include partial
derivatives and the optimization techniques that arise from them, including Lagrange

Calculus
multipliers. Volumes of rotation, arc length, and surface area are included in the additional
applications of integration. Using multiple integrals, including computing volume and center
of mass, is covered. The book concludes with an initial treatment of sequences, series, power
series, and Taylor’s series, including techniques of function approximation.

Daniel Ashlock

About SYNTHESIS
This volume is a printed version of a work that appears in the Synthesis
Digital Library of Engineering and Computer Science. Synthesis

MORGAN & CLAYPOOL

books provide concise, original presentations of important research and
development topics, published quickly, in digital and print formats.

store.morganclaypool.com
Fast Start Advanced Calculus
 Synthesis Lectures on
Mathematics and Statistics
Editor
Steven G. Kranz, Washington University, St. Louis

Fast Start Advanced Calculus

Daniel Ashlock
2019

Fast Start Integral Calculus

Daniel Ashlock
2019

Fast Start Differential Calculus

Daniel Ashlock
2019

Introduction to Statistics Using R

Mustapha Akinkunmi
2019

Inverse Obstacle Scattering with Non-Over-Determined Scattering Data

Alexander G. Ramm
2019

Analytical Techniques for Solving Nonlinear Partial Differential Equations

Daniel J. Arrigo
2019

Aspects of Differential Geometry IV

Esteban Calviño-Louzao, Eduardo García-Río, Peter Gilkey, JeongHyeong Park, and Ramón
Vázquez-Lorenzo
2019

Symmetry Problems. Thne Navier–Stokes Problem.

Alexander G. Ramm
2019
iv
An Introduction to Partial Differential Equations
Daniel J. Arrigo
2017

Numerical Integration of Space Fractional Partial Differential Equations: Vol 2 –

Applicatons from Classical Integer PDEs
Younes Salehi and William E. Schiesser
2017

Numerical Integration of Space Fractional Partial Differential Equations: Vol 1 –

Introduction to Algorithms and Computer Coding in R
Younes Salehi and William E. Schiesser
2017

Aspects of Differential Geometry III

Esteban Calviño-Louzao, Eduardo García-Río, Peter Gilkey, JeongHyeong Park, and Ramón
Vázquez-Lorenzo
2017

The Fundamentals of Analysis for Talented Freshmen

Peter M. Luthy, Guido L. Weiss, and Steven S. Xiao
2016

Aspects of Differential Geometry II

Peter Gilkey, JeongHyeong Park, Ramón Vázquez-Lorenzo
2015

Aspects of Differential Geometry I

Peter Gilkey, JeongHyeong Park, Ramón Vázquez-Lorenzo
2015

An Easy Path to Convex Analysis and Applications

Boris S. Mordukhovich and Nguyen Mau Nam
2013

Applications of Affine and Weyl Geometry

Eduardo García-Río, Peter Gilkey, Stana Nikčević, and Ramón Vázquez-Lorenzo
2013

Essentials of Applied Mathematics for Engineers and Scientists, Second Edition

Robert G. Watts
2012
 v
Chaotic Maps: Dynamics, Fractals, and Rapid Fluctuations
Goong Chen and Yu Huang
2011

Matrices in Engineering Problems

Marvin J. Tobias
2011

The Integral: A Crux for Analysis

Steven G. Krantz
2011

Statistics is Easy! Second Edition

Dennis Shasha and Manda Wilson
2010

Lectures on Financial Mathematics: Discrete Asset Pricing

Greg Anderson and Alec N. Kercheval
2010

Jordan Canonical Form: Theory and Practice

Steven H. Weintraub
2009

The Geometry of Walker Manifolds

Miguel Brozos-Vázquez, Eduardo García-Río, Peter Gilkey, Stana Nikčević, and Ramón
Vázquez-Lorenzo
2009

An Introduction to Multivariable Mathematics

Leon Simon
2008

Jordan Canonical Form: Application to Differential Equations

Steven H. Weintraub
2008

Statistics is Easy!
Dennis Shasha and Manda Wilson
2008

A Gyrovector Space Approach to Hyperbolic Geometry

Abraham Albert Ungar
2008
Copyright © 2019 by Morgan & Claypool

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means—electronic, mechanical, photocopy, recording, or any other except for brief quotations
in printed reviews, without the prior permission of the publisher.

Fast Start Advanced Calculus

Daniel Ashlock
www.morganclaypool.com

ISBN: 9781681736563 paperback

ISBN: 9781681736570 ebook
ISBN: 9781681736587 hardcover

DOI 10.2200/S00948ED1V01Y201908MAS030

A Publication in the Morgan & Claypool Publishers series

SYNTHESIS LECTURES ON MATHEMATICS AND STATISTICS

Lecture #30
Series Editor: Steven G. Kranz, Washington University, St. Louis
Series ISSN
Print 1938-1743 Electronic 1938-1751
Fast Start Advanced Calculus

Daniel Ashlock
University of Guelph

SYNTHESIS LECTURES ON MATHEMATICS AND STATISTICS #30

M
&C Morgan & cLaypool publishers
ABSTRACT
This book continues the material in two early Fast Start calculus volumes to include multivari-
ate calculus, sequences and series, and a variety of additional applications. These include partial
derivatives and the optimization techniques that arise from them, including Lagrange multipli-
ers. Volumes of rotation, arc length, and surface area are included in the additional applications
of integration. Using multiple integrals, including computing volume and center of mass, is cov-
ered. The book concludes with an initial treatment of sequences, series, power series, and Taylor’s
series, including techniques of function approximation.

KEYWORDS
partial derivatives, multivariate optimization, constrained optimization, volume,
arc-length, surface integrals, multiple integrams, series, power series
 ix

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

1 Advanced Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Implicit Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.2 Higher-order Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 The Gradient and Directional Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Tangent Planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2 Multivariate and Constrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.1 Optimization with Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2 The Extreme Value Theorem Redux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3 Lagrange Multipliers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3 Advanced Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.1 Volumes of Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2 Arc Length and Surface Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.3 Multiple Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.3.1 Mass and Center of Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

4 Sequences, Series, and Function Approximation . . . . . . . . . . . . . . . . . . . . . . . 115

4.1 Sequences and the Geometric Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.2 Series Convergence Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.2.1 Tails of Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.3 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.3.1 Using Calculus to Find Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
4.4 Taylor Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
4.4.1 Taylor Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
x
A Useful Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
A.1 Powers, Logs, and Exponentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
A.2 Trigonometric Identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
A.3 Speed of Function Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
A.4 Derivative Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
A.5 Sums and Factorization Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
A.5.1 Geometric Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
A.6 Vector Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
A.7 Polar and Rectangular Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
A.8 Integral Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
A.9 Series Convergence Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
A.10 Taylor Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

Author’s Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
 xi

Preface
This text covers multi-variable integral and differential calculus, presuming familiarity with the
single variable techniques from the precursor texts Fast Start Differential Calculus and Fast Start
Differential Calculus. The texts were developed for a course that arose from a perennial complaint
by the physics department at the University of Guelph that the introductory calculus courses
covered topics roughly a year after they were needed. In an attempt to address this concern,
a multi-disciplinary team created a two-semester integrated calculus and physics course. This
book covers the integral calculus topics from that course as well a material on the behavior of
polynomial function. The philosophy of the course was that the calculus will be delivered before
it is needed, often just in time, and that the physics will serve as a substantial collection of
motivating examples that will anchor the student’s understanding of the mathematics.
The course has run three times before this text was started, and it was used in draft form
for the fourth offering of the course, and then for two additional years. There is a good deal of
classroom experience and testing behind this text. There is also enough information to confirm
our hypothesis that the course would help students. The combined drop and flunk rate for this
course is consistently under 3%, where 20% is more typical for first-year university calculus. Co-
instruction of calculus and physics works. It is important to note that we did not achieve these
results by watering down the math. The topics covered, in two semesters, are about half again
as many as are covered by a standard first-year calculus course. That’s the big surprise: covering
more topics faster increased the average grade and reduced the failure rate. Using physics as a
knowledge anchor worked even better than we had hoped.
This text, and its two companion volumes, Fast Start Differential Calculus and Fast Start
Integral Calculus, make a number of innovations that have caused mathematical colleagues to
raise objections. In mathematics it is traditional, even dogmatic, that math be taught in an order
in which no thing is presented until the concepts on which it rests are already in hand. This is
correct, useful dogma for mathematics students. It also leads to teaching difficult proofs to stu-
dents who are still hungover from beginning-of-semester parties. This text neither emphasizes
nor neglects theory, but it does move theory away from the beginning of the course in acknowl-
edgment of the fact that this material is philosophically difficult and intellectually challenging.
The course also presents a broad integrated picture as soon as possible. The text also emphasizes
cleverness and computational efficiency. Remember that “mathematics is the art of avoiding
calculation.”
It is important to state what was sacrificed to make this course and this text work the
way they do. This is not a good text for math majors, unless they get the theoretical parts of
calculus later in a real analysis course. The text is relatively informal, almost entirely example
xii PREFACE
driven, and application motivated. The author is a math professor with a CalTech Ph.D. and
three decades of experience teaching math at all levels from 7th grade (as a volunteer) to graduate
education including having supervised a dozen successful doctoral students. The author’s calculus
credits include calculus for math and engineering, calculus for biology, calculus for business, and
multivariate and vector calculus.

Daniel Ashlock
August 2019
 xiii

Acknowledgments
This text was written for a course developed by a team including my co-developers Joanne M.
O’Meara of the Department of Physics and Lori Jones and Dan Thomas of the Department of
Chemistry, University of Guelph. Andrew McEachern, Cameron McGuinness, Jeremy Gilbert,
and Amanda Saunders have served as head TAs and instructors for the course over the last six
years and had a substantial impact on the development of both the course and this text. Martin
Williams, of the Department of Physics at Guelph, has been an able partner on the physics side
delivering the course and helping get the integration of the calculus and physics correct. I also
owe six years of students thanks for serving as the test bed for the material. Many thanks to
all these people for making it possible to decide what went into the text and what didn’t. I also
owe a great debt to Wendy Ashlock and Cameron McGuinness at Ashlock and McGuinness
Consulting for removing a large number of errors and making numerous suggestions to enhance
the clarity of the text. This is the fourth edition that corrects several mistakes and adds a very
modest number of topics.

Daniel Ashlock
August 2019
 1

CHAPTER 1

Advanced Derivatives
This chapter deals with the issues of derivatives of functions that have multiple independent
variables. In our earlier studies we learned to take derivatives of functions of the form y D f .x/.
Now we will learn to work with functions of the form z D f .x; y/, functions that graph as
surfaces in a space like the one shown in Figure 1.1.

10
y

5 10

5
x
-10 -5 5 10
(0,0,0)
origin
-5
-5

-10

-10

Figure 1.1: Three-dimensional coordinate system.

2 1. ADVANCED DERIVATIVES
Where before we had points .x; y/, we now have points .x; y; z/. We’ve looked at the graph of
functions like y D x 2 over and over. Now let’s look at the three-dimensional analog:

z D x2 C y2

for 4  x; y  4. A graph of the function is shown in Figure 1.2.

f(x,y)=sqr(x)+sqr(y):

35
30
25
20
15
10
5
0

-4
-3
-4 -2
-3 -1
-2 0
-1
0 1
1 2
2
3 3
4 4

Figure 1.2: A graph of f .x; y/ D x 2 C y 2 for 4  x; y  4.

1.1 PARTIAL DERIVATIVES

So far in this text we have had one independent and one dependent variable. Now we have two
independent variables, which has a huge effect on derivatives. The number of directions on the
number line is two – left and right or plus and minus. As we learned in Fast Start Integral Calculus,
there are an infinite number of directions when there are two variables to choose directions
among. Each unit vector starting at the origin points in a different direction. Every single vector
can be written as:
vE D .a; b/ D a
.1; 0/ C b
.0; 1/
 1.1. PARTIAL DERIVATIVES 3
In other words, each vector is a combination of the two fundamental vectors eE1 D .1; 0/ and
eE2 D .0; 1/.
On the surface of a function z D f .x; y/, the function has a rate of change in every direction at
every point. Pick a direction – the slope in that direction is the rate of change in that direction.
The basis for all of these rates of change are the partial derivatives – derivatives in the direction
of x and y .

Knowledge Box 1.1

Partial Derivatives
If z D f .x; y/ is a function of two variables, then there are two fundamental
derivatives
@f @f
zx D and zy D
@x @y
These are called the partial derivatives of z (or f ) with respect to x and y ,
respectively.

In order to take the partial derivative of a function with respect to one variable, all other variables
are treated as constants. This is most easily understood through examples.

Example 1.1 Suppose that z D x 2 C 3xy C y 2 . Find zx and zy .

Solution:

zx D 2x C 3y and zy D 3x C 2y

To see this, notice that, when we are taking the derivative with respect to x , the derivative of
3xy is 3y , because y is treated as a constant. Similarly, the derivative of 3xy with respect to y
is 3x . The derivatives for x 2 and y 2 are 2x and 2y when they are the active variable and zero
when the other variable is active, because the derivative of a constant is zero.

˙
4 1. ADVANCED DERIVATIVES
It is important to remember that every derivative rule we have learned so far applies when we are
taking partial derivatives – the product, quotient, and chain rules and all the individual formulas
for functions.
@f
Example 1.2 Find if
@x
x
f .x; y/ D
x2 C y2
Solution:

This problem requires the quotient rule. Since a “prime” hash-mark doesn’t carry its identity
@
(with respect to x or with respect to y ) we used the symbol to mean “derivative with respect
@x
to x .”

This means that

  @ @
@ x @x
x
.x 2 C y 2 / @x
.x 2 C y2/
x
D
@x x2 C y2 .x 2 C y 2 /2

1
.x 2 C y 2 / 2x
x
D
.x 2 C y 2 /2

y2 x2
D
.x 2 C y 2 /2

The next example uses the chain rule. It also uses the additional notation “fy ” as another way of
saying “the partial derivative of f .x; y/ with respect to y .”

Example 1.3 Find fy .x; y/ if

f .x; y/ D sin.2xy C 1/

Solution:
fy .x; y/ D cos.2xy C 1/
2x

˙
 1.1. PARTIAL DERIVATIVES 5
As long as we remember that y is the active variable and x is treated as a constant, this is not
difficult.

Example 1.4 Find the partial derivatives with respect to x and y of

f .x; y/ D .3x C 4y/5

Solution:
fx .x; y/ D 5.3x C 4y/4
3 fy .x; y/ D 5.3x C 4y/4
4
The only point where things are different is the way the chain rule acts depending on if x or y
is the active variable.
˙

Example 1.5 Find the partial derivatives with respect to x and y for
x
f .x; y/ D
y

Solution:
1 x
fx .x; y/ D fy .x; y/ D
y y2

1
For fx , is effectively a constant, while, for fy , we employ the reciprocal rule while x plays the
y
part of a constant.
˙

1.1.1 IMPLICIT PARTIAL DERIVATIVES

Since natural laws are often stated in the form of equations that are not in functional form,
implicit derivatives are very useful in physics. It turns out that implicit partial derivatives are a
lot like standard partial derivatives – as long as you remember which variable is active.

Example 1.6 Find zx and zy if x 2 C y 2 C z 2 D 16.

Solution:

z is the dependent variable and so gets a zx or a zy each time we take a derivative of it,
while x and y take turns being the active variable and a constant, respectively. So,

2x C 2z
zx D 0 and 2y C 2z
zy D 0
6 1. ADVANCED DERIVATIVES
Simplifying we get that
x y
zx D and zy D
z z
˙

Example 1.7 Find zy if

.xyz C 1/3 D 6y
Solution:

In this example z is the dependent variable, y is the active variable, and x is acting like
a constant. So:

3.xyz C 1/2 .xz C xy
zy / D 6

2
xz C xy
zy D
.xyz C 1/2
2 z
zy D 2
xy
.xyz C 1/ y

1.1.2 HIGHER-ORDER PARTIAL DERIVATIVES

In earlier chapters, when we wanted the second derivative of a function we just took the deriva-
tive again. The fact that we have multiple choices of which derivative to take complicates this.
If the first derivative is with respect to x or y and so is the second, then we get four possible
nominal second derivatives:

fxx fxy fyx fyy

A useful fact saves us from having the number of higher-order partial derivatives explode.
Knowledge Box 1.2

When working with f .x; y/,

fxy .x; y/ D fyx .x; y/
and in general the order in which partial derivatives are taken does not affect the
result of taking them.
 1.1. PARTIAL DERIVATIVES 7
Example 1.8 If

f .x; y/ D x 3 C 3xy C y 2

find fxx , fxy , fyx , and fyy , and verify that fxy D fyx .

Solution:

Start by finding fx and fy and then keep going.

fx D 3x 2 C 3y

fy D 3x C 2y

fxx D 6x

fxy D 3

fyx D 3

fyy D 2

And we see that fxy D 3 D fyx , achieving the desired verification. From this point on we will
only compute one of the two mixed partials fxy and fyx .
˙

The “order doesn’t matter” rule means, for example, that fxxy D fxyx D fyxx . So the degree to
which this rule reduces the number of higher-order derivatives increases with the order of the
derivative.

Example 1.9 Find fxx , fxy , and fyy for

f .x; y/ D x 2 sin.y/:
8 1. ADVANCED DERIVATIVES
Solution:
Compute the first partials first and keep going.

fx D 2x sin.y/

fy D x 2 cos.y/

fxx D 2 sin.y/

fxy D 2x cos.y/

fyy D x 2 sin.y/

Example 1.10 Find fxy for

xy
f .x; y/ D :
x2C1
Solution:

The fact that we get the same result by computing the partials with respect to x and y in
either order means that we may choose the order to minimize our work. The order y then x is
easier, because y vanishes.
x
f .x; y/ D y

x2 C1

x
fy D
x2 C1

.x 2 C 1/.1/ x.2x/
fxy D 2
Quotient rule.
.x 2 C 1/

1 x2
D
.x 2 C 1/2

˙
 1.1. PARTIAL DERIVATIVES 9
PROBLEMS
Problem 1.11 For each of the following functions find fx and fy .

1
2 2
1. f .x; y/ D 2x C 3xy C y C 4x C 2y C 7 5. s.x; y/ D
x2
C y2 C 1
2. g.x; y/ D sin.xy/ x y
6. q.x; y/ D
xCy
3. h.x; y/ D x 3 y 3
7. a.x; y/ D ex sin.y/
4. r.x; y/ D ln.x 2 C y 2 C 1/
8. b.x; y/ D .x 2 C xy C 1/6

Problem 1.12 For each of the following functions find fxx , fxy , and fyy .

1
1. f .x; y/ D x 2 5xy C 2y 2 C 3x 6y C 11
5. s.x; y/ D
x2
C y2 C 1
2. g.x; y/ D tan 1 .xy/ 2x 3y
6. q.x; y/ D
3. h.x; y/ D .x 3 C 1/.y 3 C 1/ 5x C y
2 Cy 2 7. a.x; y/ D ln.x sin.y//
4. r.x; y/ D ex 5

8. b.x; y/ D .2x 2 xy/4

Problem 1.13 Find zx and zy if .xyz C 2/3 D 4.

x
Problem 1.14 Find zx and zy if D 1.
yz

xCz
Problem 1.15 Find zx and zy if D 4.
yCz

3x C z
Problem 1.16 Find zx and zy if D 4xy .
2z
p
2
Problem 1.17 Find zx and zy if cos.x C y C z/ D .
2

Problem 1.18 Find zx and zy if tan 1 .z xy/ D 1.

Problem 1.19 Find zx and zy if .xy C xz C yz/3 D 16.

10 1. ADVANCED DERIVATIVES
Problem 1.20 For each of the following functions find fxy .

1. f .x; y/ D x 2 y C x 3 y C y sin.x/ 5. s.x; y/ D x
sin.xy/

2. g.x; y/ D x sin.y/ C x tan 1 .x/ 6. q.x; y/ D y
cos.xy/

x y
3. h.x; y/ D xy 2 C x 2 y C xy 7. a.x; y/ D C C y 3
y x

2
4. r.x; y/ D x.y C 1/5 C 1 8. b.x; y/ D .1 C x C y C xy/4

x2
Problem 1.21 Find fx , fy , fxx , fxy , and fyy if f .x; y/ D .
x2 C y2

Problem 1.22 Find gx , gy , gxx , gxy , and gyy if g.x; y/ D tan 1 .xy C 1/.

Problem 1.23 Rind fxx , fxxy , and fxxyy if h.x; y/ D sin.xy/.

Problem 1.24 Find zx and zy if z D .x 2 C 1/y .

Problem 1.25 Find zx and zy if z xy D 2.

x 3 .x C 1/2 .x 1/3
Problem 1.26 Find zx and zy if z D .
y 2 .y 1/3 .y C 1/5

1.2 THE GRADIENT AND DIRECTIONAL DERIVATIVES

We are now ready to look at some of the opportunities that are available once we understand
partial derivatives. At any point on the surface that forms the graph of z D f .x; y/ there are
an infinite number of directions and so an infinite number of rates at which the function is
changing. Pick a direction, and the function has a rate of change in that direction.
It turns out that there is some order to this richness of directions and rates of growth, in the
form of a simple formula for the direction in which the function is growing fastest.
 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 11
Knowledge Box 1.3

If z D f .x; y/ is a function of two variables, then

rf .x; y/ D fx ; fy
is called the gradient of f .x; y/. The gradient of a function points in the direction
it is growing most quickly; the rate of growth is the magnitude of the gradient.

Example 1.27 Find the gradient of the function f .x; y/ D x 2 C y 2 C 3xy .

Solution:

Using the formula given, rf .x; y/ D .2x C 3y; 3x C 2y/.

We can ask much more complex questions about the gradient than simply computing its value.
Example 1.28 At what points is the function
g.x; y/ D sin.x/ C cos.y/
changing the fastest in its direction of maximum increase?

Solution:

This question wants us to maximize the magnitude of the gradient.

First compute the gradient:

rg.x; y/ D .cos.x/; sin.y//
The magnitude of this is q
cos2 .x/ C sin2 .y/
Since x and y vary independently, the answer is simply those points that make cos2 .x/ and
sin2 .y/ both one.

So, the answer is those points .x; y/ such that

2m C 1
x D n and y D 
2
12 1. ADVANCED DERIVATIVES
where n and m are whole numbers.
˙

The graph in Figure 1.3 might help you understand Example 1.28.

f(x,y)=sin(x)+cos(y)

2
1.5
1
0.5
0
-0.5
-1
-1.5
-2

8
6
4
2
-8 0
-6
-4 -2
-2
0 -4
2 -6
4
6
8 -8

Figure 1.3: A graph of z D sin.x/ C cos.y/ for 8  x; y  8.

Example 1.29 Find the gradient of

z D xy
Solution:

The partial derivative with respect to x is not hard because y is treated as a constant – so
zx D yx y 1 . The partial derivative with respect to y is trickier. It uses the formula for the
derivative of a constant to a variable power which gives zy D x y
ln.x/.

This makes the gradient

rz D yx y 1
; x y
ln.x/
˙
 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 13
What is the physical meaning of the gradient? We have already noted that it points in the
direction in which a surface grows fastest away from the point where the gradient is computed
– the steepest uphill slope away from the point. The magnitude of the gradient also gives of the
rate of fastest growth. Another fact is that the negative of the function

rf .x; y/

is the steepest downhill slope away from the point.

In other words, the negative of the gradient is the direction that a ball, starting at rest,
will roll.

f(x,y)=2xy/(x^2+y^2+1)

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1

5
4
3
2
-3 1
-2
-1 0
0
1 -1
2 -2
3
4
5 -3

2x C y
Figure 1.4: A graph of H.x; y/ D for 3  x; y  5
x2 C y2 C 1

2x C y
Example 1.30 Suppose that the function H.x; y/ D (Figure 1.4) describes the
x2
C y2 C 1
height of a surface. Which direction will a ball placed at the point .1; 2/ roll?
14 1. ADVANCED DERIVATIVES
Solution:

We need to compute the negative of the gradient of H.x; y/.

.x 2 C y 2 C 1/.2/ .2x C y/.2x/

Hx .x; y/ D
.x 2 C y 2 C 1/2

12 8
Hx .1; 2/ D D 1=9
36

.x 2 C y 2 C 1/.1/ .2x C y/.2y/

Hy .x; y/ D
.x 2 C y 2 C 1/2

6 8
Hy .1; 2/ D D 1=18
36

So the ball rolls in the direction of the vector

 
1 1
rH.1; 2/ D ;
9 18
˙

Notice that, if we were designing a game, then we could use a well-chosen equation to give us
a height map for the surface, and the gradient could be used to tell which ways balls would roll
and water would flow.
Visualization helps us understand functions. This leads to the question: what does a gradient
look like? The gradient of a function f .x; y/ assigns a vector to each point in space. That means
that we could get an idea of what a gradient looks like by plotting the vectors of the gradient on
a grid of points.

Example 1.31 Let f .x; y/ D x 2 C y 2 . For all points .x; y/ with coordinates in the
set f˙2; ˙1:5; ˙1; ˙0:5; 0g, plot the point and the vector starting at that point in the direction
rf .x; y/.

Solution:

Since rf .x; y/ D .2x; 2y/ the vectors are:

 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 15

2.5

-2.5 2.5

-2.5

Notice that all the gradient vectors point directly away from the origin. This corresponds to our
notion that the gradient is the direction of fastest growth.

2x C y
The function from Example 1.30, the rolling ball question, will probably have a
x2C y2 C 1
more interesting gradient than the simple paraboloid in Example 1.31.

2x C y
Example 1.32 Let H.x; y/ D . For all points .x; y/ with coordinates in the set
x2 C y2 C 1
f˙3; ˙2:5; ˙2; ˙1:5; ˙1; ˙0:5; 0g, plot the point and the vector starting at that point in the
direction rH.x; y/.
16 1. ADVANCED DERIVATIVES
Solution:

We are plotting the vectors drawn from the gradient

 2 
2y 2x 2 2xy C 2 x 2 y 2 4xy C 1
rH.x; y/ D ;
.x 2 C y 2 C 1/2 .x 2 C y 2 C 1/2

which yields the picture:

3.5

-3.5 3.5

-3.5

The gradient vectors point in many directions. A ball rolling on this surface could potentially
have a very complex path.
˙
 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 17
Example 1.33 Find a reasonable sketch of the vector field associated with the gradient
of g.x; y/ D sin.x/ C cos.y/ from Example 1.28. Use grid points with coordinates ˙n for
n D 0; 1; : : : 8.

Solution:
We are plotting the vectors drawn from the gradient

rg.x; y/ D .cos.x/; sin.y//

which yields the picture:

9.0

-9.0 9.0

-9.0

The periodicity of the vector field is easy to see. The different cells of the periodicity are slightly
different because the periods are multiples of  , and the grid used to display the vectors is sized
in multiples of one.
˙
18 1. ADVANCED DERIVATIVES
Now that we have the gradient, it is possible to define the derivative in a particular direction.
Knowledge Box 1.4

If z D f .x; y/ is a differentiable function of two variables, and uE D .a; b/ is a

unit vector, then the derivative of f .x; y/ in the direction of uE is:
ruE f .x; y/ D uE
rf .x; y/ D afx C bfy

If vE D .r; s/ is any vector, then the derivative of f .x; y/ in the direction of vE

is:
vE
rvE f .x; y/ D
rf .x; y/
jE
vj

Notice that we are continuing the practice of using unit vectors to designate directions – even
when computing the derivative of a function in the direction of a general vector, we first coerce
it to be a unit vector.
It is worth mentioning that, when computing directional derivatives, we start with a scalar
quantity – the function f .x; y/. When we compute the gradient, we get the vector quantity
rf .x; y/ D .fx ; fy / but then return to a scalar function of two variables ruE f .x; y/. It is im-
portant to keep track of the type of object – scalar or vector – that you are working with.

Example 1.34 Find the derivative of

f .x; y/ D x 2 C y 2
p
in the direction of uE D .1=2; 3=2/.

Solution:

The vector uE is a unit vector so, starting with fx D 2x , fy D 2y , we get:

p
1 3 p
ruE f .x; y/ D 2x C 2y D x C 3y
2 2
˙

The directional derivative occasionally comes up in the natural course of trying to solve a prob-
lem. There is one very natural application: finding level curves. First let’s define level curves.
 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 19
Knowledge Box 1.5

Level Curves
If z D f .x; y/ defines a surface, then the level curve of height c of f .x; y/ is
the set of points that solve the equation
f .x; y/ D c:

Example 1.35 Plot the level curves for c 2 f1; 2; 3; 4; 5; 6g for

f .x; y/ D x 2 C 2y 2

Solution:
p
The equation x 2 C 2y 2 D c is an ellipse that is 2 times as far across in the x direction
as the y directions. Solving for the points where x D 0 or y D 0 gives us the extreme points of
the ellipse for each value of c , and we get the following picture.

2.0
c=6

c=2
c=1
-3.0 3.0

-2.0
Level curves for f .x; y/ D x 2 C 2y 2 .

Since the values of c used to derive the level curves are equally spaced, the fact that the curves are
getting closer together gives a sense of how steeply the graph of f .x; y/ D x 2 C 2y 2 is sloped.
˙
20 1. ADVANCED DERIVATIVES
In Example 1.35 we can plot the level curves because we recognize them as ellipses. In
general, that’s not going to work. If we supply a graphics system with information about the
gradient of a curve, then it can sketch level curves by following the correct directional derivative.

Knowledge Box 1.6

If z D f .x; y/ defines a surface, then the level curve at the point .a; b/ proceeds
in the direction of the unit vector uE such that
jruE f .a; b/j D 0:
In other words, it proceeds in the direction such that the height of the graph of
f .x; y/ is not changing.

Example 1.36 Find, in general, the direction of the level curves of

f .x; y/ D x 2 C 3y 2 :

Solution:

To find the general direction of level curves, we need to solve this equation for uE .

jruE f .a; b/j D 0

Let uE D .a; b/:

r x 2 C 3y 2
.a; b/ D 0

.2x; 6y/
.a; b/ D 0

2xa C 6yb D 0

2xa D 6yb
y 
aD 3 b
x

Giving us the direction in which the level curves go.

 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 21
If we set b D 1, then, at a given point .x; y/ in space, the level curve at that point is in
3y
the direction ; 1 . This encodes the direction.
x

To be thorough, let’s turn this into a unit vector.

ˇ s r
ˇ  2
ˇ
ˇ 3y ˇ 3y 9y 2 1p 2
ˇ ; 1 ˇˇ D 2
C1 D 2
C1D 9y C x 2
x x x x

So, the direction, as a unit vector is:

!
3y x
p ;p
9y 2 C x 2 9y 2 C x 2

Notice that if we had chosen b D 1, we would have gotten

!
3y x
p ; p
9y 2 C x 2 9y 2 C x 2

which is an equally valid solution.

The level curve points in two opposite directions.

˙
22 1. ADVANCED DERIVATIVES
A natural question at this point is: what do the vectors we found in Example 1.36 look like?

Example 1.37 Using the solution to Example 1.36 plot the directions of the level curves as a
vector field.

Solution:
4.0

-4.0 4.0

-4.0

At each point, there are actually two vectors that point in the direction of the level curve. If uE
points in the direction of the level curve, then so does uE . The figure above sometimes shows
one and sometimes the other. The curves that these vectors are following track the vertically long
ellipses that one would expect as level curves of:

f .x; y/ D x 2 C 3y 2

˙
 1.2. THE GRADIENT AND DIRECTIONAL DERIVATIVES 23
PROBLEMS

Problem 1.38 For each of the following functions, find their gradient.

1. f .x; y/ D x 2 C 4xy C 7y 2 C 2x 5y C 1 5. s.x; y/ D sin.x/ C cos.y/

2. g.x; y/ D sin.xy/ 6. q.x; y/ D x 5 y C xy 5 C x 3 y 3 C 1
x2 x
3. h.x; y/ D 2 7. a.x; y/ D 2
y C1 x C y2

3=2 8. b.x; y/ D x y
4. r.x; y/ D x 2 C y 2

Problem 1.39 For the function

g.x; y/ D x 2 C 3xy C y 2 ;

remembering that directions should be reported as unit vectors, find:

1. The greatest rate of growth of the curve in any direction at the point .1; 1/
2. The direction of greatest growth at .1; 1/
3. The greatest rate of growth of the curve in any direction at the point . 1; 2/
4. The direction of greatest growth at . 1; 2/
5. The greatest rate of growth of the curve in any direction at the point .0; 3/
6. The direction of greatest growth at .0; 3/

Problem 1.40 Suppose we are trying to find level curves. Is it possible to find points where
there are more than two directions in which the surface does not grow? Either explain why this
cannot happen or give an example where it does.

Problem 1.41 For each of the following functions, sketch the gradient vector field of the func-
tion for the points .x; y/ with

x; y 2 f0; ˙1; ˙2; ˙3g:

Exclude points, if any, where the gradient does not exist. This is a problem where a spreadsheet
may be useful for performing routine computation.
24 1. ADVANCED DERIVATIVES
p
1. f .x; y/ D x 2 C y 2 C 1 4. r.x; y/ D x2 C y2
2. g.x; y/ D sin.xy/ 5. s.x; y/ D sin.x/ C cos.y/
x
3. h.x; y/ D
xCy 6. q.x; y/ D x 5 y C xy 5 C x 3 y 3 C 1

Problem 1.42 Find the directional derivative of

f .x; y/ D x 2 C y 3

in the direction of (1,1) at (0,0).

Problem 1.43 Find the directional derivative of

g.x; y/ D x 2 y2
p
in the direction of . 3=2; 1=2/ at (2,1).

Problem 1.44 Find the directional derivative of

h.x; y/ D .y 2x/3

in the direction of (0,-1) at (-1,1).

Problem 1.45 As in Example 1.36 find the general direction for level curves at .x; y/ for the
function
1
H.x; y/ D
x2 C y2
and plot the vector field for all points with whole number coordinates in the range 3  x; y 
3.

Problem 1.46 As in Example 1.36 find the general direction for level curves at .x; y/ for the
function
2
H.x; y/ D
x2 2x C y 2 2y C 2
and plot the vector field for all points with whole number coordinates in the range 5  x; y 
5.

Problem 1.47 Which way will a ball placed at (1,1) on the surface given by the function in
Problem 1.42 roll?
 1.3. TANGENT PLANES 25
1.3 TANGENT PLANES
One of the first things we built after developing skill with the derivative was tangent lines to
a curve. With a function z D f .x; y/ the analogous object is a tangent plane. There are several
ways to specify a plane.

Knowledge Box 1.7

Formulas for planes

If a, b , c , and d are constants, all of the following formulas specify planes:
z D ax C by C c
ax C by C cz D d:
.a; b; c/
.x; y; z/ D d
Note that the third formula, while phrased in terms of a dot product, is actually
the same as the second.

Let’s get some practice with converting between the different possible forms of a plane.

Example 1.48 If
.3; 2; 5/
.x; y; z/ D 2
find the other two forms of the plane.

Solution:

.3; 2; 5/
.x; y; z/ D 2
3x 2y C 5z D 2 Second form
5z D 3x C 2y C 2
z D 0:6x C 0:4y C 0:4 First form

The most common way to find a tangent line for a function is to take the point of tangency –
which must be on the line – together with the slope of the line found by computing the derivative
and use the point slope form to find the formula for the line. It turns out that there is a way of
specifying planes that is similar to the point slope form of a line. Remember that if vE and wE are
vectors that are at right angles to one another, then vE
wE D 0.
26 1. ADVANCED DERIVATIVES
Knowledge Box 1.8

Formula for a plane at right angles to a vector

Suppose that vE is a vector at right angles to a plane in three dimensions and that
.a; b; c/ is a point on that plane. Then a formula for the plane is
vE
.x a; y b; z c/ D 0:
Notice that .a; b; c/ is constructively on the plane and that our knowledge of the
dot product tells us it is at right angles to vE.

Example 1.49 Find the plane at right angles to vE D .1; 2; 1/ through the point .2; 1; 5/.

Solution:

Simply substitute into the formula given in Knowledge Box 1.8.

.1; 2; 1/
.x 2; y C 1; z 5/ D 0
1.x 2/ C 2.y C 1/ C 1.z 5/ D 0
x C 2y C z 2 C 2 5 D 0
x C 2y C z D 5

Knowledge Box 1.8 seems very special purpose, but it turns out to be very useful in light of
another fact. Suppose that
f .x; y; z/ D c

specifies a surface. We need to expand our definition of the gradient just a bit to


rf .x; y; z/ D fx .x; y; z/; fy .x; y; z/; fz .x; y; z/ :

In this case the vector rf .a; b; c/ points directly outward from the surface f .x; y; z/ D c at
.a; b; c/ and it is at right angles to the tangent plane.
 1.3. TANGENT PLANES 27
Knowledge Box 1.9

Formula for the tangent plane to a surface

If f .x; y; z/ D c defines a surface, and .a; b; c/ is a point on the surface, then a
formula for the tangent plane to that surface at .a; b; c/ is:


fx .a; b; c/; fy .a; b; c/; fz .a; b; c/
.x a; y b; z c/ D 0:

Defining a surface in the form f .x; y; z/ D c is a little bit new – but in fact this is another
version of level curves, just one dimension higher. Let’s practice.

Example 1.50 Find the tangent plane to the surface x 2 C y 2 C z 2 D 3 at the point .1; 1; 1/.

Solution:

Check that the point is on the surface: .1/2 C . 1/2 C .1/2 D 3 – so it is. Next find the
gradient
rx 2 C y 2 C z 2 D .2x; 2y; 2z/:

This means that the gradient at .1; 1; 1/ is vE D .2; 2; 2/. This makes the plane

.2; 2; 2/
.x 1; y C 1; z 1/ D 0
2x 2y C 2z 2 2 2 D 0
2x 2y C 2z D 6
x yCz D3

Notice that we simplified the form of the plane; this is not required but it does make for neater
answers.

The problem with finding the tangent plane to a surface f .x; y; z/ D c is that it does not solve the
original problem – finding tangent planes to z D f .x; y/. A modest amount of algebra solves
this problem. If z D f .x; y/ then g.x; y; z/ D z f .x; y/ D 0 is in the correct form for our
surface techniques. This gives us a new way of finding tangent planes to a function that defines
a surface in 3-space.
28 1. ADVANCED DERIVATIVES
Knowledge Box 1.10

Formula for the tangent plane to a functional surface

If z D f .x; y/ defines a surface, then the tangent plane to the surface at .a; b/
may be obtained as


fx .a; b/; fy .a; b/; 1
.x a; y b; z f .a; b// D 0:
This is the result of applying the gradient-of-a-surface formula to the surface z
f .x; y/ D 0 at the point .a; b; f .a; b//.

Example 1.51 Find the tangent plane to f .x; y/ D x 2 y 3 at the point .2; 1/.

Solution:
Assemble the pieces and plug into Knowledge Box 1.10.

fx .x; y/ D 2x

fx .2; 1/ D 4

fy .x; y/ D 3y 2

fy .2; 1/ D 3

f .2; 1/ D 4 . 1/ D 5

Put the plane together

. 4; 3; 1/
.x 2; y C 1; z 5/ D 0

4x C 8 C 3y C 3 C z 5D0

4x C 3y C z D 6

Which is the tangent plane desired.

˙
 1.3. TANGENT PLANES 29
Example 1.52 Find the tangent plane to

g.x; y/ D xy 2

at (3,1).

Solution:

Assemble the pieces and plug into Knowledge Box 1.10.

fx .x; y/ D y 2

fx .3; 1/ D 1

fy .x; y/ D 2xy

fy .3; 1/ D 6

f .3; 1/ D 3

Put the plane together

. 1; 6; 1/
.x 3; y 1; z 3/ D 0

xC3 6y C 6 C z 3D0

x 6y C z D 6

PROBLEMS

Problem 1.53 Find the plane through (1,-1,1) at right angles to vE D .2; 2; 1/.

Problem 1.54 Find the plane through (2,0,5) at right angles to vE D .1; 1; 1/.

Problem 1.55 Find the plane through (3,2,1) at right angles to vE D .1; 1; 2/.
30 1. ADVANCED DERIVATIVES
Problem 1.56 Find, in the form

ax C by C cz D d;

the tangent planes to the following curves at the indicated points.

1. f .x; y/ D x 2 C y 2 1 at .2; 2/ 4. r.x; y/ D y
ln.x 2 C 1/ at .3; 1/

2 Cy 2
2. g.x; y/ D 3x 2 C 2xy C 42 1 at . 1; 1/ 5. s.x; y/ D ex at .0; 1/
3. h.x; y/ D .x C y C 1/2 at .0; 4/ 6. q.x; y/ D sin.x/ cos.y/ at .=3; =6/

Problem 1.57 Suppose that

x 2 C y 2 C z 2 D 12:

Find the tangent plane at each of the following points.

p
1. p D .2; 2; 2/ 3. r D . 2; 2; 2/ 5. v D . 7; 1; 2/
p p
2. q D .2; 2; 2/ 4. u D .1; 1; 10/ 6. w D . 1=2; 2:5; 22=2/

Problem 1.58 Find all tangent planes to

x 2 C y 2 C z 2 D 75

that are at right angles to the vector .2; 4; 6/.

Problem 1.59 Find the tangent plane at . 1; =3/ to

g.x; y/ D x cos.y/
 1.3. TANGENT PLANES 31
Problem 1.60 Find the tangent plane to each of the following surfaces at the indicated point.

1. Surface x C y C z 2 D 6 at (1,1,2) 5. Surface xyz C x C y C z D 4 at (1,1,1)

2. Surface x 2 y 3 C 5z D 19 at (4,-2,-1) 6. Surface xy C yz C xz D 0 at (-1,1,-2)

3. Surface 3x C y 2 C z 2 D 8 at (2,1,1) 7. Surface x 2 C y 2 C z 3 D 6 at (2,1,1)
4. Surface .x y/3 C 2z D 14 at (2,0,3) 8. Surface xy C xz C yz D 7 at (1,1,3)

Problem 1.61 Find the tangent plane at .0; 0/ to

.x 2 Cy 2 /
g.x; y/ D e

Problem 1.62 Find the tangent plane at .=4; =4; =4/ to

cos.xyz/ D 0

Problem 1.63 If
x2 C y2 C z2 D r 2
is a sphere of radius r centered at the origin .0; 0; 0/ (it is), show that a sphere has a tangent
plane at right angles to any non-zero vector.

Problem 1.64 If you want to find the tangent plane to a point on a sphere, what is the simplest
method? Explain.

Problem 1.65 If P and Q are planes that are both at right angles to a vector vE, and they are
not equal, what can be said about the intersection of P and Q?

Problem 1.66 Suppose that uE and vE are vectors so that vE
uE D 0. If P is a plane at right angles
to uE , and Q is a plane at right angles to vE in three-dimensional space, then what is the most that
can be said about the intersection of the planes?
 33

CHAPTER 2

Multivariate and Constrained

Optimization

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

4
3
2
1
-4 0
-3
-2 -1
-1
0 -2
1 -3
2
3
4 -4

1
Figure 2.1: A graph of f .x; y/ D on 4  x; y  4.
x2 C y2 C 1

A big application of derivatives in our earlier work was optimizing functions. In this chapter we
learn to do this for multiple independent variables. In some ways we will be doing pretty much
the same thing – but as always, more variables makes the process more complicated.

2.1 OPTIMIZATION WITH PARTIAL DERIVATIVES

Examine the hill-shaped graph in Figure 2.1. This function has a single optimum at
.x; y/ D .0; 0/. It is also positive everywhere. It is a good, simple function for demonstration
34 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
purposes. The level curves are circles centered on the origin – except at the origin on top of the
optimum, where we get a “level curve” consisting of a single point. This sort of point – where
the directional derivatives are all zero – is the three-dimensional equivalent of a critical point.

Knowledge Box 2.1

Critical points for surfaces

If z D f .x; y/ defines a surface, then the critical points of the function are at
the points .x; y/ that solve the equations:
fx .x; y/ D 0 and fy .x; y/ D 0:

As with our original optimization techniques, local and global optima occur at
critical points or at the boundaries of the domain of optimization.

The information in Knowledge Box 2.1 gives us a good start on finding optima of multivariate
functions – but it also contains a land mine. The phrase the boundaries of the domain of optimization
is more fearsome when there are more dimensions. The next section of this chapter is about
dealing with some of the kinds of boundaries that arise when optimizing surfaces.

1
Example 2.1 Demonstrate that the function in Figure 2.1, f .x; y/ D , in fact
x2 C y2 C 1
has a critical point at (0,0) by solving for the partial derivatives equal to zero.

Solution:
The partial derivatives are:

2x 2y
fx .x; y/ D 2
and fy .x; y/ D
.x 2 C y 2 C 1/ .x 2 C y 2 C 1/2

Remembering that a fraction is zero only when its numerator is zero – and noting the denomi-
nators of these partial derivative are never zero – we see we are solving the very difficult system
of simultaneous equations:
2x D 0 and 2y D 0

So we verify a single critical point at (0,0).

˙
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 35
Example 2.2 Find the critical point(s) of

g.x; y/ D x 2 C 2y 2 C 4xy 6x 8y C 2

Solution:

Start by computing the partials.

gx .x; y/ D 2x C 4y 6
gy .x; y/ D 4y C 4x 8

This gives us the simultaneous system:

2x C 4y 6D0
4x C 4y 8D0

2x C 4y D6
4x C 4y D8
2x D2 Second line minus first.
x D1
2 C 4y D6 Plug x=1 into first line.
4y D4
y D1

So we find a single critical point at .x; y/ D .1; 1/.

Now that we can locate critical points of surfaces, we have to deal with figuring out if the point is
a local maximum, a local minimum, or something else. Let’s begin by understanding the option
of “something else.” Examine the function in Figure 2.2. The partial derivatives are fx D 2x
and fy D 2y . So, it has a critical point at (0,0) in the center of the graph, but it does not have an
optimum. This is a type of critical point called a saddle point.
36 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION

100
80
60
40
20
0
-20
-40
-60
-80
-100

10
5
-10 0
-5
0 -5
5
10-10

Figure 2.2: A graph of f .x; y/ D x 2 y 2 on 10  x; y  10. This graph exemplifies a saddle

point.

Knowledge Box 2.2

Types of critical points for surfaces

When we find critical points by solving fx D 0 and fy D 0, there are three pos-
sible outcomes:
• if all nearby points are lower, the point is a local maximum,
• if all nearby points are higher, the point is a local minimum,
• if there are nearby points that are both higher and lower, the point is a sad-
dle point.

If we have a good graph of a function, then we can often look at a critical point to determine
its character. This is not always practical. The sign chart technique from Fast Start Differential
Calculus is not possible – it only makes sense in a one-dimensional setting. There is an analog to
the second derivative test. Look at the functions in Figures 2.1 and 2.2. If we slice them along
the x and y axes, we see that the hill in Figure 2.1 is concave down along both axes, while the
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 37
saddle in Figure 2.2 is concave down along the y axis and concave up along the x axis. This type
of observation can be generalized into a type of second derivative test.

Knowledge Box 2.3

Classifying critical points for surfaces

For a function z D f .x; y/ that defines a surface, define
2
D.x; y/ D fxx .x; y/fyy .x; y/ fxy .x; y/:

The quantity D is called the discriminant of the system. Then for a critical point
.a; b/
• If D.a; b/ > 0 and fxx .a; b/ > 0 or fyy .a; b/ > 0, then the point is a local
minimum.
• If D.a; b/ > 0 and fxx .a; b/ < 0 or fyy .a; b/ < 0, then the point is a local
maximum.
• If D.a; b/ < 0, then the point is a saddle point.
• If D.a; b/ D 0, then the test yields no information about the type of the
critical point.

The fourth outcome – no information – forces you to examine the function in other ways. This
situation is usually the result of having functions with repeated roots or high powers of both
variables. The function

q.x; y/ D x 3 C y 3 ;

for example, has a critical point at .0; 0/, but all of fxx , fyy , and fx;y are zero. The critical
point is a saddle point, but that conclusion follows from careful examination of the graph of
the function, not from the second derivative test.

A part of the structure of this multivariate version of the second derivative test is that either fxx
or fyy may be used to check if a critical point is a minimum or maximum. A frequent question
by students is “which should I use?” The answer is “whichever is easier.” The reason for this is
that the outcome of the discriminant test that tells you a critical point is an optimum of some
sort forces the signs of fxx and fyy to agree.
38 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Examine the formula for D.a; b/ and see if you can tell why this is true.

Example 2.3 Find and classify the critical point of

h.x; y/ D 4 C 3x C 5y 3x 2 C xy 2y 2

Solution:

The partials are fx D 3 6x C y and fy D 5 C x 4y . Setting these equal to zero we

obtain the simultaneous system of equations:

6x y D 3
x C 4y D 5

Solve:
6x C 24y D 30

23y D 33

y D 33=23

x C 132=23 D 115=23

xD 17=23

x D 17=23

Now compute D.x; y/. We see fxx D 6, fyy D 4 and fxy D 1. So,

D D . 6/. 4/ 12 D 23:

Since D > 0, the critical point is an optimum of some sort. The fact that fxx < 0 or the fact
that fyy < 0 suffices to tell us this point is a maximum.

Notice that in Example 2.3 the equation is a quadratic with large negative squared terms and
a small mixed term (xy ). The fact that its sole critical point is a maximum means that it is a
paraboloid opening downward. In the next example we look at a quadratic with a large mixed
term.
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 39
Example 2.4 Find and classify the critical points of

q.x; y/ D x 2 C 6xy C y 2 14x 10y C 3:

Solution:

Start by computing the needed partials and the discriminant.

fx .x; y/ D 2x C 6y 14

fy .x; y/ D 6x C 2y 10

fxx D 2

fyy D 2

fxy D 6

D.x; y/ D 2
2 62 D 32

These calculations show that any critical point we find will be a saddle point since D < 0. To
find the critical point we solve the first partials equal to zero obtaining the system

2x C 6y D 14

6x C 2y D 10

6x C 18y D 42 3x line one

16y D 32 Line 3 minus line two

yD2

2x C 12 D 14 Substitute

2x D 2

xD1

So we see that the function has, as its sole critical point, a saddle point at (1,2).
˙
40 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
The last two examples have been bivariate quadratic equations with a constant discriminant. This
is a feature of all bivariate quadratics. All functions of this kind have a single critical point and
a constant discriminant. For the next example we will tackle a more challenging example.

Example 2.5 Find and classify the critical points of f .x; y/ D x 4 C y 4 16xy C 6:

Solution:

Start by computing the needed partials and the discriminant.

fx .x; y/ D 4x 3 16y

fy .x; y/ D 4y 3 16x

fxx .x; y/ D 12x 2

fyy .x; y/ D 12y 2

fxy .x; y/ D 16

2
D.x; y/ D fxx fyy fxy

D.x; y/ D 144x 2 y 2 256

Next, we need to find the critical points by solving the system

4x 3 D 16y

4y 3 D 16x

or

x 3 D 4y

y 3 D 4x
p
3
Solve the second equation to get y D 4x and plug into the first, obtaining:
p
3
x 3 D 4 4x

x 9 D 256x Cube both sides

 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 41
9
x 256x D 0

x.x 8 256/ D 0
p
8
x D 0; ˙ 256

x D 0; ˙2

Since the equation system is symmetric in x and y we may deduce that y D 0; ˙2 as well.
Referring back to the original equations, it is not hard to see that, when x D 0, y D 0; when
x D 2 so must y ; when x D 2 so must y. This gives us three critical points: . 2; 2/; .0; 0/;
and .2; 2/. Next, we check the discriminant at the critical points.

D. 2; 2/ D 2304 256 D 2048 > 0

Since fxx . 2; 2/ > 0, this point is a minimum.

D.2; 2/ D 2048 > 0

Another minimum, and

D.0; 0/ D 256;
so this point is a saddle point.
˙

One application of multivariate optimization is to minimize distances. For that we should state,
or re-state, the definition of distance.

Knowledge Box 2.4

The definition of distance

If p D .x0 ; y0 / and q D .x1 ; y1 / are points in the plane, then the distance be-
tween p and q is
p
d.p; q/ D .x0 x1 /2 C .y0 y1 /2 :

If r D .x0 ; y0 ; z0 / and s D .x1 ; y1 ; z1 / are points in space, then the distance

between r and s is
p
d.r; s/ D .x0 x1 /2 C .y0 y1 /2 C .z0 z1 /2 :
42 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
The definition in Knowledge Box 2.4 can be extended to any number of dimensions – the dis-
tance between two points is the square root of the sum of the squared differences of the individual
coordinates of the point. For this text only two- and three-dimensional distances are required.
With the definition of distance in place, we can now pose a standard type of problem.

Example 2.6 What point on the plane z D 3x C 2y C 4 is closest to the origin?

Solution:

The origin is the point .0; 0; 0/, while a general point .x; y; z/ on the plane has the form
.x; y; 3x C 2y C 4/. That means that the distance we are trying to minimize is given by
p
dD .x 0/2 C .y 0/2 C .3x C 2y C 4 0/2

p
D 10x 2 C 12xy C 5y 2 C 24x C 16y C 16

The partial derivatives of this function are:

20x C 12y C 24
dx D p
2 10x 2 C 12xy C 5y 2 C 24x C 16y C 16

12x C 10y C 16
dy D p
2 10x C 12xy C 5y 2 C 24x C 16y C 16
2

We need to solve the simultaneous equation in which each of these partials is zero. Remember a
fraction is zero only where its numerator is zero. This maxim gives us the simultaneous equations:

20x C 12y C 24 D 0

12x C 10y C 16 D 0

Simplify

5x C 3y D 6

6x C 5y D 8
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 43
Solve:
30x C 18y D 36

30x C 25y D 40

7y D 4

yD 4=7

5x 12=7 D 42=7

5x D 30=7

xD 6=7

So we find a single critical point . 6=7; 4=7/ Considering the geometry of a plane, we see
that it has a unique closest approach to the origin. So, computing z D 3x C 2y C 4 D 18=7
8=7 C 28=7 D 2=7, gives us that the point on the plane closest to the origin is
. 6=7; 4=7; 2=7/:
˙

Clever students will have noticed that when we were minimizing

p
10x 2 C 12xy C 5y 2 C 24x C 128 C 16
the numerators of the partial derivatives were exactly the partial derivatives of
10x 2 C 12xy C 5y 2 C 24x C 128 C 16. This observation is an instance of a more general
shortcut.

Knowledge Box 2.5

Minimization of distance—a shortcut

p
Suppose
p that we are optimizing a function g.x; y/ D f .x; y/ or g.x; y; z/ D
f .x; y; z/. Then the critical points for optimization of g and f are the same,
and the second derivative tests agree. The actual values of the function are not,
which means care is required.
44 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
p p
Why does this shortcut work? It is because if 0 < a < b , then 0 < a < b . So optimizing the
value finds where the optimum of the square root of the value is as well.

Example 2.7 Minimize: p

g.x; y/ D x2 C y2 C 3
Solution:
p
In this case g.x; y/ D f .x; y/ where f .x; y/ D x 2 C y 2 C 3, so the shortcut applies.
Finding the relevant partials we see that

fx .x; y/ D 2x
fy .x; y/ D 2y

which is easy to see has a critical point at (0,0). The second derivative test shows that:
2
fxx fyy fxy D2
2 0D4>0

So the critical point is an optimum. Since fp xx D 2 > 0, it is a minimum. This means the
minimum value of g.x; y/ is at the point .0; 0; 3/.

Definition 2.1 A function m.x/ is monotone increasing if, whenever a < b and m.x/ exists on
Œ a; b , then m.a/ < m.b/.
p
Notice that m.x/ D x is monotone increasing. In fact the shortcut in Knowledge Box 2.5
works for any monotone increasing function. These functions include: ex , ln.x/, tan 1 .x/, x n
p
when n is odd, and n x .
Knowledge Box 2.6

A test for a function being monotone increasing

We know that a function is increasing if its first derivative is positive. If a func-
tion exists and has a positive derivative on the interval [a,b] it is a monotone
increasing function on [a,b].
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 45
Example 2.8 Show that y D ln.x/ is monotone increasing where it exists.

Solution:
According to Knowledge Box 2.6, a function is increasing when its first derivative is positive.
The function y D ln.x/ only exists on the interval .0; 1/. Its derivative is y 0 D 1=x which is
positive for any positive x . Thus, ln.x/ is increasing everywhere that it exists and so is a monotone
increasing function.
˙
Using an extension of the shortcut from Knowledge Box 2.5 with these other monotone func-
tions will make the homework problems much easier. Let’s practice.

Example 2.9 Find the point of closest approach to the origin on the plane

3x 4y z D 4:

Solution:

First find the distance between the origin .0; 0; 0/ and a generic point .x; y; 3x 4y 4/
on the plane. We get that
p
d.x; y/ D x 2 C y 2 C 9x 2 24xy C 16y 2 32y 24x C 16
p
d.x; y/ D 10x 2 24xy C 17y 2 32y 24x C 16
Using the distance minimization shortcut, find the critical points of

d 2 .x; y/ D 10x 2 24xy C 17y 2 32y 24x C 16

which is the relatively simple quadratic case.

fx .x; y/ D 20x 24y 24

fy .x; y/ D 24x C 34y 32

fxx .x; y/ D 20

fyy .x; y/ D 34

fx;y .x; y/ D 24

D.x; y/ D 20
34 . 24/2 D 104

46 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
So we see that D > 0 and fxx > 0 meaning we will find a minimum distance. Now we must
solve for the critical point.

20x 24y D 24

24x C 34y D 32

5x 6y D 6

12x C 17y D 16

60x 72y D 72

60x C 85y D 80

13y D 152

y D 152=13

5x 912=13 D 78=13

5x D 990=13

x D 198=13

So the critical point is roughly at x D 15:23, y D 11:69, making the point of closest approach
.15:23; 11:69; 2:93/. Not a lot easier – but we avoided all sorts of square roots.

PROBLEMS
Problem 2.10 Find the critical points for each of the following functions, and use the second
derivative test to classify them as maxima, minima, or saddle points. Also, find the value of the
function at the critical point.

1. f .x; y/ D x 2 C y 2 6x 4y C 4 4. q.x; y/ D x 2 C xy C y 2 C 4x 5y C 2
2 3 2 2 4 4
2. g.x; y/ D 3x y C y 3x 3y C 2 5. r.x; y/ D x C y 4xy C 1
2 2 3 3
3. h.x; y/ D 12 x C 3xy y C 2x y C 1 6. s.x; y/ D x C 2y 4xy
 2.1. OPTIMIZATION WITH PARTIAL DERIVATIVES 47
Problem 2.11 Show that
f .x; y/ D x 4 C y 4
is an example of a function where the second derivative test yields no useful information.

Problem 2.12 Suppose that

f .x; y/ D ax 2 C bxy C cy 2 C dx C ey C f:

Show that this function has at most one critical point. When it does have a critical point, derive
rules based on the constants a-f for classifying that critical point.

Problem 2.13 Find the point on the plane

xCyCz D4

closest to the origin.

Problem 2.14 Find the point on the plane

zD x=3 C y=4 C 1

closest to the origin.

Problem 2.15 Find the point on the tangent plane of

f .x; y/ D x 2 C y 2 C 1

at .2; 1; 6/ that is closest to the origin.

Problem 2.16 Find the critical points for each of the following functions, and use the second
derivative test to classify them as maxima, minima, or saddle points. Also, find the value of the
function at the critical point.

1. f .x; y/ D ln.x 2 C y 2 C 4/ 4. r.x; y/ D .x 2 C xy C 3y 2 4x C 3y C 1/5

p
2. g.x; y/ D ex
4 Cy 4 36xyC6 5. s.x; y/ D x 2 C y 2 C 14
 2 2

3. h.x; y/ D tan 1 .x 3 C 2y 3 4xy/ 6. q.x; y/ D tan 1 ex CxyCy 2x 3yC1
48 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Problem 2.17 For each of the following functions either demonstrate it is not monotone in-
creasing on the given interval or show that it is.

1. y D ex on . 1; 1/ x2
5. y D on .0; 1/
x2 C 1
2. y D x 3 x on . 1; 1/
2x
x 6. y D on . 1; 1/
3. y D 2 on . 1; 1/ x2C1
x C1
ex
1
4. y D tan .x/ on . 1; 1/ 7. y D on . 1; 1/
ex C 1

Problem 2.18 Suppose that we are going to cut a line of length L into three pieces. Find the
division into pieces that maximizes the sum of the squares of the lengths. This can be phrased
as a multivariate optimization problem.

Problem 2.19 Find the largest interval on which the function

x
x2 C 16
is monotone increasing.

Problem 2.20 Find the largest interval on which the function

x2 1
x2 C 4
is monotone increasing.

2.2 THE EXTREME VALUE THEOREM REDUX

The extreme value theorem, last seen in Fast Start Differential Calculus, says that The global max-
imum and minimum of a continuous, differentiable function must occur at critical points or at the
boundaries of the domain where optimization is taking place. This is just as true for optimizing a
function z D f .x; y/. So what’s changed? The largest change is that the boundary can now have
a very complex shape.

In the initial section of this chapter we carefully avoided optimizing functions with boundaries,
thus avoiding the issue with boundaries. In the next couple of examples we will demonstrate
techniques for dealing with boundaries.
 2.2. THE EXTREME VALUE THEOREM REDUX 49
Example 2.21 Find the global maximum of
x y
f .x; y/ D .xy C 1/e

for x; y  0.

Solution:

This function is to be optimized only over the first quadrant. We begin by finding criti-
cal points in the usual fashion
x y x y x y
fx .x; y/ D ye C .xy C 1/e . 1/ D . xy C y 1/e

x y x y x y
fy .x; y/ D xe C .xy C 1/e . 1/ D . xy C x 1/e

Since the extreme value theorem tells us that the optima occur at boundaries or critical points,
we won’t need a second derivative test – we just compare values. This means out next step is to
solve for any critical points. Remember that powers of e cannot be zero, giving us the system of
equations:

xy C y 1D0

xy C x 1D0

xy D x 1Dy 1 so x D y

x2 C x 1D0

x2 xC1D0
p
1˙ 1 4
xD There are no critical points!
2
This means that the extreme value occurs on the boundaries – the positive x and y axes where
either x D 0 or y D 0. So, we need the largest value of f .0; y/ D e y for y  0 and f .x; 0/ D
e x for x  0. Since e x is largest (for non-negative x ) at x D 0, we get that the global maximum
is f .0; 0/ D 1.
˙

While the extreme value theorem lets us make decisions without the second derivative test, it
forces us to examine the boundaries, which can be hard. It is sometimes possible to solve the
50 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
problem by adopting a different point of view. The next example will use a transformation to a
parametric curve to solve the problem.

Example 2.22 Maximize

g.x; y/ D x 4 C y 4

for those points f.x; y/ W x 2 C y 2  4g.

Solution:

This curve has a single critical point at .x; y/ D .0; 0/ which is extremely easy to find.
The boundary for the optimization domain is the circle x 2 C y 2 D 4, a circle of radius 2
centered at the origin. This boundary is also the parametric curve .2 cos.t/; 2 sin.t//. This means
that on the boundary the function is

g.2 cos.t /; 2 sin.t // D 16 cos4 .t/ C 16 sin4 .t/:

This means we can treat the location of optima on the boundary as a single-variable optimization
task of the function g.t / D 16 cos4 .t / C 16 sin4 .t/. We see that

g 0 .t / D 64 cos3 .t/
. sin.t// C 64 sin3 .t / cos.t / D 64 sin.t/ cos.t/.sin2 .t/ cos2 .t//

Solve:

sin.t/ D 0 t D .2n C 1/
2

cos.t/ D 0 t D n

sin2 .t / cos2 .t/ D 0


sin2 .t/ D cos2 .t/ t D ˙.2n C 1/
4
If we plug these values of t back into the parametric
p pcurve, the points on the boundary
that may be optima are: .0; ˙2/, .˙2; 0/, and .˙ 2; ˙ 2/. Plugging the first four of these
into the curvepwe get that g.x; y/ D 16. Plugging the last four into the function we get
g.x; y/ D 2
. 2/4 D 2
4 D 8. At the critical point (0,0) we see g.x; y/ D 0. This means that
the maximum value of g.x; y/ on the optimization domain is 16 at any of .0; ˙2/; .˙2; 0/.

˙
 2.2. THE EXTREME VALUE THEOREM REDUX 51
Notice that the solutions to the parametric version of the boundary gave us an infinite number
of solutions. But, when we returned to the .x; y; z/ domain, this infinite collection of solutions
were just the eight candidate points repeated over and over. This means that our failure to put a
bound on the parameter t was not a problem; bounding t was not necessary.
A problem with the techniques developed in this section is that all of them are special purpose.
We will develop general purpose techniques in Section 2.3 – the understanding of which is
substantially aided by the practice we got in this section.

Example 2.23 If
h.x; y/ D x 2 C y 2

find the minimum value of h.x; y/ among those points .x; y/ on the line y D 2x 4.

Solution:

The function h.x; y/ is a paraboloid that opens upward. If we look at the points .x; y; z/
on h.x; y/ that happen to lie on a line, that line will slice a parabolic shape out of the surface
defined by h.x; y/. First note that h.x; y/ has a single critical point at .0; 0/ which is not in
the domain of optimization. This means we may consider only those points on y D 2x 4, in
other words on the function

h.x; 2x 4/ D x 2 C .2x 4/2 D 5x 2 16x C 16

This means that the problem consists of finding the minimum of f .x/ D 5x 2 16x C 16.

f 0 .x/ D 10x 16 D 0

10x D 16

x D 8=5

y D 16=5 20=5 D 4=5

Since f .x/ opens upward, we see that this point is a minimum; there are no boundaries to the
line that is constraining the values of .x; y/ so the minimum value of h.x; y/ on the line is

64 16 80 16
h.8=5; 4=5/ D C D D
25 25 25 5

˙
52 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Example 2.23 did not really use the extreme value theorem. For that to happen we would need
to optimize over a line segment instead of a full line.

Example 2.24 If
h.x; y/ D x 2 C y 2

find the minimum and maximum value of h.x; y/ among those points .x; y/ on the line
y D x C 1 for 4  x  4.

Solution:

This problem is very similar to Example 2.23. The ends of the line segment are (-4,-3)
and (4,5), found by substituting into the formula for the line. On the line h.x; y/ becomes

h.x; x C 1/ D x 2 C .x C 1/2 D 2x 2 C 2x C 1

So we get a critical point at 4x C 2 D 0 or x D 1=2 which is the point . 1=2; 1=2/.

Plug in and we get

h. 4; 3/ D 25

h. 1=2; 1=2/ D 1=2

h.4; 5/ D 41

This means the minimum value is 1/2 at the critical point and that the maximum value is 41 at
one of the endpoints of the domain of optimization.

Making the domain a line segment is one of the simplest possible options. Let’s look at another
example with a more complex domain of optimization.

Example 2.25 If
s.x; y/ D 2x C y 4

find the minimum and maximum value of s.x; y/ among those points .x; y/ on the curve
y D x 2 3 for 2  x  2.
 2.2. THE EXTREME VALUE THEOREM REDUX 53
Solution:

In this problem we are cutting a parabolic segment out of the plane s.x; y/ D 2x C y 4. We
get that the ends of the domain of optimization are .2; 1/ and . 2; 1/ by plugging in the ends of
the interval in x to the formula for the parabolic segment. Substituting the parabolic segment
into the plane yields

s.x; x 2 3/ D 2x C x 2 3 4 D x 2 C 2x 7

This means our critical point appears at 2x C 2 D 0 or x D 1, making the candidate point
. 1; 2/. Plugging the candidate points into s.x; y/ yields:

s. 2; 1/ D 7

s. 1; 2/ D 8

s.2; 1/ D 1

This means that the maximum value is 1 at .2; 1/, and the minimum is 8 at . 1; 2/.

Some sets of boundaries are simple enough that we can use geometric reasoning to avoid needing
to use calculus on the boundaries.

Example 2.26 Find the maximum value of

h.x; y/ D x 2 C y 2

for 2  x; y  3.

Solution:

First of all, we know that this surface has a single critical point at .0; 0/ – this surface is
an old friend (see graph in Figure 1.2).

The domain of optimization is a square with corners . 2; 2/ and .3; 3/. Along each side
of the square, we see that the boundary is a line – and so has extreme values at its ends. This
means that we need only check the corners of the square; the interior of the edges – viewed as
line segments – cannot attain maximum or minimum values by the extreme value theorem. This
means we need only add the points . 2; 2/, . 2; 3/, .3; 2/, and .3; 3/ to our candidates.
54 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION

Plugging in the candidate points we obtain:

h. 2 2/ D 8

h.0; 0/ D 0

h. 2; 3/ D 13

h.3; 2/ D 13

h.3; 3/ D 18

So the maximum is 18 at .3; 3/, and the minimum is 0 at .0; 0/.

The goal for this section was to set up LaGrange Multipliers – the topic of Section 2.3. The
take-home message from this section is that the extreme value theorem implies that optimizing
on a boundary is the difficult added portion of optimizing on a bounded domain.
The techniques that we will develop in the next section require that the boundary itself be a
differentiable curve. Some of the boundaries in this section are made of several differentiable
curves, meaning that the techniques in this section may be easier for those problems. If we have
a boundary that is not differentiable, then the techniques in this section are the only option.

PROBLEMS

Problem 2.27 If we look at the points on

h.x; y/ D x 2 C y 2 C 4x C 4

that fall on a line, then there is a minimum somewhere on the line. Find that minimum value
for the following lines.

p
1. y D 3x C 1 3. x C y D 6 5. x C y D 3=2

2. y D 4 x 4. 2x 7y D 24 6. 3x C 5x D 7
 2.2. THE EXTREME VALUE THEOREM REDUX 55
Problem 2.28 Find the maximum of

f .x; y/ D .x 2 C y 2 / e xy

with .x; y/ in the first quadrant where 0 < x; y .

Hint: this function is symmetric. Use this fact.

Problem 2.29 Find the maximum and minimum of

1
q.x; y/ D
x2 C y2 2x 4y C 6

in the first quadrant: 0 < x; y .

Problem 2.30 Find the maximum of

g.x; y/ D .x 4 C y 4 /

on the set of points

f.x; y/ W x 2 C y 2  25g:

Problem 2.31 Maximize

h.x/ D x 4 C y 4

on the set of points f.x; y/ W x 2 C y 2  25g.

Problem 2.32 If
g.x; y/ D 2x 2 C 3y 2 ;

find the minimum of g.x; y/ for those points .x; y/ on the line y D x 1.
56 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Problem 2.33 For the function

s.x; y/ D x 2 C y 2 2x C 4y C 1

with .x; y/ on the following line segments, find the minimum and maximum values.

1. y D 2x 2 on 2  x  2 4. x C y D 10 on 4  x  10
2. y D x C 7 on 1  x  4 5. 2x 7y D 8 on 10  x  10

3. y D 6 5x on 0  x  6 6. 2x y D 13 on 1  x  5

Problem 2.34 Suppose that

p.x; y/ D 3x 5y C 2
Find the maximum and minimum values on the following parametric curves.

1. .3t C 1; 5 t/ I 5t 5 4. .sin.2t/; cos.t //

2. .cos.t /; sin.t// 5. .t cos.t/; t sin.t// I 0  t  2

3. .sin.t/; 3 cos.t // 6. .3 sin.t/; 2 sin.t//

Problem 2.35 Find the maximum and minimum of

1
q.x; y/ D
x2 C y2 C 4x 12y C 45

in the first quadrant: 0 < x; y .

Problem 2.36 Suppose that P D f .x; y/ is a plane and that we are considering the points
where x 2 C y 2 D r 2 for some constant r . If f .x; y/ is not equal to a constant, explain why
there is a unique minimum and a unique maximum value.

Problem 2.37 If
f .x; y/ D x 2 C y 2 ;
find the minimum of f .x; y/ for those points .x; y/ on the line y D mx C b .
 2.3. LAGRANGE MULTIPLIERS 57
2.3 LAGRANGE MULTIPLIERS

This section introduces constrained optimization using a technique called Lagrange mul-
tipliers. The basic idea is that we want to optimize a function f .x; y/ at those points where
g.x; y/ D c . The function g.x; y/ is called the constraint.

The proof that Lagrange multipliers work is beyond the scope of this text, so we will begin by
just stating the technique.

Knowledge Box 2.7

The method of LaGrange Multipliers with two variables

Suppose that z D f .x; y/ defines a surface and that g.x; y/ D c spec-
ifies points of interest. Then the optima of f .x; y/, subject to the con-
straint that g.x; y/ D c , occur at solutions to the system of equations
fx .x; y/ D 
gx .x; y/

fy .x; y/ D 
gy .x; y/

g.x; y/ D c
where  is an auxiliary variable.

The variable  is new and strange – it is the “multiplier” – and, as we will see, correct solutions
to the system of equations that arise from Lagrange multipliers typically use  in a fashion that
causes it to drop out.

If the constraint g.x; y/ D c is thought of as the boundary of the domain of optimization, then
using Lagrange multipliers gives us a tool for resolving the boundary as a source of optima as
per the extreme value theorem.

With that context, let’s practice our Lagrange multipliers.

58 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Example 2.38 Find the maxima and minima of

f .x; y/ D x C 4y 2

on the ellipse 2x 2 C 3y 2 D 36.

Solution:

The equations arising from the Lagrange multiplier technique are:

1 D 
4x

4 D 
6y

2x 2 C 3y 2 D 36

Solving:

1
xD
4

2
yD
3

Plug into the constraint and we get:

2 12
C D 36
162 9
2

1 4
C D 362
8 3

35=24 D 362

35=864 D 2

or
p
 D ˙ 35=864
 2.3. LAGRANGE MULTIPLIERS 59
Which yields candidate points:
r r
1 864 54
xD˙ D˙
4 35 35
r r
2 864 384
yD˙ D˙
3 35 35

Since the equation of f .x; y/ is a plane that gets larger as x and y get larger, we see that the
maximum is
p p 
f 54=35; 384=35 Š 12:5

and the minimum is

 p p 
f 54=35; 384=35 Š 16:5

Notice that in the calculations in Example 2.38, the auxiliary variable  served as an informa-
tional conduit that let us discover the candidate points. The next example is much simpler.

Example 2.39 Find the point of closest approach of the line 2x C 5y D 3 to the origin (0,0).

Solution:

This problem is like the closest approach of a plane to the origin problems, but in a lower dimen-
sion. The tricky part of this problem is phrasing it as a function to be minimized and a constraint.

Since we are minimizing distance we get that the function is the distance of a point
.x; y/ from the origin:
p p
d.x; y/ D .x 0/2 .y 0/2 D x2 C y2

As per the monotone function shortcut, we can instead optimize the function

d 2 .x; y/ D x 2 C y 2 :

The constraint function is the line. We can phrase the line as a constraint by saying:

g.x; y/ D 2x C 5y D 3
60 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
With the parts in place, we can extract the Lagrange multiplier equations.

2x D 2

2y D 5

2x C 5y D 3

Solve:

xD

5
yD

2
 
5
2./ C 5
 D3
2

29

D3
2

6
D
29

6 15
Yielding: gx D and y D
29 29
 
6 15
This makes the point on 2x C 5y D 3 closest to the origin ; .
29 29
˙

Example 2.40 Find the maximum value of

f .x; y/ D x 2 C y 3

subject to the constraint x 2 C y 2 D 9.

 2.3. LAGRANGE MULTIPLIERS 61
Solution:

This problem is already in the correct form for Lagrange multipliers, so we may immedi-
ately derive the system of equations.
2x D 2x

2y D 3y 2 

x2 C y2 D 9

The first equation yields the useful information that  D 1 but that x may take on any value,
unless x D 0 in which case  may be anything.

Given that  D 1, the second equation tells us y D 0 or y D 2=3. If  is free and x D 0

then y may be anything.

Plugging the values for y into the constraint that says the p
points lie on a circle,
p we can retrieve
values for x : when y D 0, x D ˙3; when y D 2=3, x D ˙ 9 4=9 D ˙ 77=9.

If x D 0 and y is free the constraint yields y D ˙3.

p p
This gives us the candidate points .˙3; 0/, .0; ˙3/, .2=3; 77=3/, and .2=3; 77=3/.
The sign of the x -coordinate is unimportant because f .x; y/ depends on x 2 ; since f .x; y/
depends on y 3 , positive values yield larger values of f , and negative ones yield smaller values
of f .
p
Since f .3; 0/ D 9, f .2=3; 77=3/ Š 25:47, and f .0; 3/ D 27, we get that the maximum value
of the function on the circle is:
f .0; 3/ D 27

In Section 2.2 we found the minimum of a line on a quadratic surface that opens upward (Ex-
ample 2.23). The next example lets us try a problem like this using the formalism of Lagrange
multipliers.
Example 2.41 Find the minimum of f .x; y/ D x 2 xy C y 2 on the line 5x C 7y D 18.
62 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Solution:

The constraint is the line – so g.x; y/ D 5x C 7y D 18. With this detail we can apply
the Lagrange multiplier technique:

2x y D 5

2y x D 7

5x C 7y D 18

18 5x
Solve the constraint for y and we get y D so
7

18 5x
2x D 5
7

14x 18 C 5x D 35

19x 35 D 18

and
 
18 5x
2 x D 7
7

36 10x 7x D 49

17x C 49 D 36

D .17x C 36/=49

19x C 35
.17x C 36/=49 D 18

1526x D 378

xD 189=763 Š 0:25

y D 2097=763 Š 2:75
 2.3. LAGRANGE MULTIPLIERS 63
Making the minimum

 2  2
189 189
2097 2097
C Š 8:30
763 742 763

This problem actually got harder when we used the Lagrange multiplier formalism to solve it.
This is another example of how different tools are good for different problems.

At this point, we introduce the 3-space version of Lagrange multipliers. Or, we could
say that this is the version of Lagrange multipliers that uses three independent variables. This
version of the Lagrange multiplier technique widens the variety of problems we can work with.

Knowledge Box 2.8

The method of LaGrange Multipliers with three variables
Suppose that w D f .x; y; z/ defines a surface and that g.x; y; z/ D c specifies
points of interest. Then the optima of f .x; y; z/, subject to the constraint that
g.x; y; z/ D c , occur at solutions to the system of equations
fx .x; y; z/ D 
gx .x; y; z/

fy .x; y; z/ D 
gy .x; y; z/

fz .x; y; z/ D 
gz .x; y; z/

g.x; y; z/ D c
where  is an auxiliary variable.

Example 2.42 Find the point of intersection of the function w D 2x C y z and the sphere
x 2 C y 2 C z 2 D 8 that has the largest value of w .
64 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Solution:

The problem is already in the correct form to apply Lagrange multipliers.

2x D 2

2y D 

2z D 

x2 C y2 C z2 D 8

Solving the first three equations tell us that

1 1
xD yD  zD 
2 2

Plugging these into the last equation tells us that

2 C .1=4/2 C .1=4/2 D 8

.3=2/2 D 8

2 D 16=3
p
 D ˙4= 3

4 2 2
So we see that: x D ˙ p y D ˙p z D ˙p
3 3 3
Looking at the formula for w , we see that w grows as x , y , and z .
 
4 2 2
So the point that maximizes w is: .x; y; z/ D p ; p ; p :
3 3 3
˙

Example 2.43 Suppose we divide a rope of length 6 m into three pieces. What size of pieces
maximizes the product of the lengths?
 2.3. LAGRANGE MULTIPLIERS 65
Solution:

Name the length of the pieces x , y , z . That means the function we are maximizing is
f .x; y; z/ D xyz and the constraint is x C y C z D 6.

Having put the problem into the form for Lagrange multipliers, we can move to the system of
equations.

yz D 

xz D 

xy D 

xCyCz D6

So x D =z D y and z D =x D y making x D y D z . Since they sum to 6, we see

x D y D z D 2 is the sole candidate point. Testing constrained points near .2; 2; 2/, like
.1:9; 2:1; 2/ and .2; 1:95; 2:05/, shows that this point is a maximum.

PROBLEMS

Problem 2.44 For each of the following sets of functions and constraints, write out but do not
solve the Lagrange multiplier equations.

1. f .x; y/ D cos.x/ sin.y/ constrained by x 2 C y 2 D 8

2. g.x; y/ D xe y constrained by 2x 3y D 12

x2
3. h.x; y/ D constrained by 3x 2 C y 2 D 48
y2 C1

4. r.x; y/ D 2xy constrained by x 2 C y 2 D 8

5. s.x; y/ D ln.x 2 C y 2 C 4/ constrained by xy D 9

6. q.x; y/ D tan 1 .xy C 1/ constrained by x 2 y2 D 4

66 2. MULTIVARIATE AND CONSTRAINED OPTIMIZATION
Problem 2.45 For each of the following lines, find the point on the line closest to the origin
using the method of Lagrange multipliers.

1. y D 2x C 1 3. 2x C 3y D 17 5. y D 5x C 25
2. y D 4 x 4. x C y D 12 6. 2x C 4y D 8

Problem 2.46 Using Lagrange multipliers, find the maximum and minimum values of

f .x; y/ D x 4 C y 4

subject to the constraint that

x 2 C y 2 D 1:

Problem 2.47 Suppose that

z D 3x 5y C 2
For each of the following constraints, find the maximum and minimum values of z subject to
the constraint, if any.

1. x 2 C y 2 D 4 4. x 2 xy C y 2 D 1

2. x 2 C y 2 D 0:25 5. x 2 y2 D 4

3. 2x 2 C 6y 2 D 64 6. xy D 16

Problem 2.48 If
f .x; y/ D x 3 y2
and x 2 C y 2 D 25, find the maximum and minimum values f .x; y/ can take on.

Problem 2.49 Use Lagrange multipliers to find the closest approach of the plane

f .x; y/ D 3x yC2

to the origin.
 2.3. LAGRANGE MULTIPLIERS 67
Problem 2.50 If
f .x; y; z/ D x 2 C 4y 2 C 9z 2
and x C y C z D 60, find the values of x , y , and z that maximize and minimize f .

Problem 2.51 If
h.x; y; z/ D x 2 y2 C z2
and x C y C z D 6, find the values of x , y , and z that maximize and minimize h.

Problem 2.52 If
q.x; y; z/ D x 2 y2 2z 2
and x C y C z D 300, find the values of x , y , and z that maximize and minimize q .

Problem 2.53 Find the largest point in intersection of

x C 2y C 3z D 4

and a cylinder of radius 2 centered on the z -axis.

 69

CHAPTER 3

Advanced Integration
This chapter covers various sorts of integration that compute volumes and areas. The first, vol-
umes of revolution, is a small twist on the integration methods we have already studied. The
other techniques involve multivariate integration, which is both newer and more difficult. This
latter subject permits us to compute the volume under a surface as we computed the area under
the curve earlier.

3.1 VOLUMES OF ROTATION

Figure 3.1: A portion of a curve rotated about the x -axis.

70 3. ADVANCED INTEGRATION

Figure 3.2: Approximations of a rotated volume using smaller and larger numbers of disks.

Figure 3.1 shows a portion of a curve and its shadow reflected about the x -axis. This is meant
to evoke revolving the curve about the axis between the two vertical lines. When the curve is
rotated, it encloses a volume – our goal is to compute that volume.
When integration was defined in Fast Start Integral Calculus, we first approximated the area
represented by the integral with rectangles and then let the number of rectangles go to infinity,
obtaining both the area and the integral as a limit. For volumes of rotation, we use a similar tech-
nique with disks instead of rectangles. Figure 3.2 shows how a conical volume is approximated
using five, ten, and twenty disks. The more disks we use, the closer the sum of the volumes of
the disks gets to the volume enclosed by rotating a curve about the x axis.
The final version of the area integral added up functional heights – rectangles of infinite thinness.
Recall that the area under a curve from x D a to x D b was

Z b
Area D f .x/
dx:
a

Given this, the analogous integral for volume of rotation is

Z b Z b
Volume D Area of Circles
dx D  r 2
dx
a a

The radius of the circular slices of the volume we are trying to compute is given by the height
of the function being rotated about the axis, giving us the final formula for volume of rotation of
a function about the x -axis.
 3.1. VOLUMES OF ROTATION 71
Knowledge Box 3.1

Formula for volume of rotation about the x axis

The volume V enclosed by rotating the function f .x/ about the x -axis from
x D a to x D b is: Z b
V D f .x/2
dx:
a

(Notice that we took the constant  out in front of the integral sign.)

Example 3.1 Find the volume enclosed by rotating y D x 2 about the x axis from x D 0 to
x D 2.

Solution:

Start by sketching the situation:

(2,4)

(0,0)
x
(2,0)
72 3. ADVANCED INTEGRATION
Using the equation in Knowledge Box 3.1 we get:
Z 2
V D  f .x/2
dx
0
Z 2
2
D  x2
dx
0
Z 2
D  x 4
dx
0
ˇ
1 5 ˇˇ2
D  x ˇ
5 0
 
32
D  0
5
32
D units3
5
And so we see the volume of the broad trumpet-shaped solid enclosed by rotating y D x 2 about
the x -axis from x D 0 to x D 2 has a volume of 32=5 Š 503 units3 .
˙

The next example is similar except that it uses a more difficult integral. If you’re not comfortable
with integration by parts, please review it in Fast Start Integral Calculus.

Example 3.2 Find the volume enclosed by rotating y D ln.x/ about the x axis from x D 1 to
x D 4.
Solution:

Start again with a sketch.

y

ln.x/
(4,ln.4/)

x
(1,0)
 3.1. VOLUMES OF ROTATION 73
Applying the formula from Knowledge Box 3.1 we get:
Z 4
Volume D  ln.x/2
dx
1
U D ln.x/2 V Dx
 Z 4 
2 ln.x/
D  x ln.x/2 2 ln.x/dx dU D
dx d V D dx
1 x

U D ln.x/ V Dx
 Z 4 
dx
D  x ln.x/2 2x ln.x/ C 2 dx dU D dV D dx
1 x
ˇ4
ˇ
D .x ln.x/ 2
2x ln.x/ C 2x/ˇˇ
1

D 4 ln.4/2 8 ln.4/ C 8 0C0 2

Š 8:16 units3

Now that we have a formula for rotating objects about the x -axis, the next logical step is to
rotate them about the y -axis. This turns out to be a little trickier.

There are two basic techniques.

1. We can figure out a new function (the inverse of the original one) that gives us disks along
the y -axis, and integrate with respect to dy , or

2. We can use a different type of slice, the cylindrical shell. Instead of slicing the shape into
disks and integrating them, we slice it into cylinders.

The radius of the disks for rotation about the x -axis was simply the value of the function, f .x/,
but the distance from a graph to the y axis is just x . We need to get the information about the
y -distance in somehow. We will start with Method 2, the method that uses a different type of
slice (cylinders centered on the y -axis). Figure 3.3 shows the result of rotating a line about the
y axis.
The surface area of a cylinder, not counting the top and bottom, is 2 rh where r is its radius
and h is its height. The radius, as already noted, is x , and the height of the cylinder is y D f .x/.
74 3. ADVANCED INTEGRATION
y

Figure 3.3: Cylindrical shells for computing a volume of revolution about the y -axis.

Thus, we have the area of our cylindrical slices

Area D 2
r
h D 2
x
f .x/

Knowledge Box 3.2

Volume of rotation with cylindrical shells about the y -axis

If we rotate a function f .x/ about the y -axis, then the volume enclosed by the
curve between x D a and x D b is:
Z xDb
V D 2 x
f .x/
dx:
xDa

(Notice that we took the constant 2 out in front of the integral sign.)

One tricky thing about this is that the limits of integration are the range of x -values that the
radii of the cylindrical shells span. The y -distances only come in via the participation of f .x/.
The formula in Knowledge Box 3.2 assumes we are rotating the area below the curve around the
y axis. Finding other areas may require a more complicated setup where we need to figure out
the height of the cylinders.
 3.1. VOLUMES OF ROTATION 75
Example 3.3 Find the volume of rotation of the area below the curve f .x/ D 1=x about the
y -axis from x D 0:5 to x D 2:0.

Solution:

Sketch the situation.

x
2

Applying the equation in Knowledge Box 3.2 we get:

Z 2
Volume D 2 x
f .x/
dx
0:5
Z 2
1
D 2 x

dx
0:5 x
Z 2
D 2 dx
0:5
ˇ2
ˇ
D 2x ˇˇ
0:5

D 2.2 0:5/

D 3 units3

˙
76 3. ADVANCED INTEGRATION
In the next example we will try and find the volume enclosed by rotating a curve about the y
axis. This means we will need to be much more careful about the heights of the cylinders.
Example 3.4 Find the volume enclosed by rotating the area bounded by f .x/ D x 2=3 , x D 1,
and y D 32=3 about the y -axis from x D 1 to x D 3.

Solution:

Start with the traditional sketch.

y

.3; 32=3 )

y D x 2=3

.1; 1/

x
x=1 x=3

This example is one where the height of the cylinder is not just f .x/, so we need to find an
expression for the height of the cylinder. This expression goes from f .x/ (at the bottom) to
y D 32=3 (at the top). So, we get a formula of
h D 32=3 f .x/ D 32=3 x 2=3


for the height of the cylinder. The area of a cylinder is thus 2x
32=3 x 2=3 which makes the
volume:
Z 3  
Volume D 2 x
32=3 x 2=3 dx
1
Z 3  
D 2 x
32=3 x
x 2=3 dx
1
Z 3  
D 2 32=3 x x 5=3 dx
1
!ˇ3
32=3 2 3 8=3 ˇ
ˇ
D 2 x x ˇ
2 8 ˇ
1
Š 10:52 units3
˙
 3.1. VOLUMES OF ROTATION 77
Example 3.4 demonstrates that the formula in Knowledge Box 3.2 only covers one type of rota-
tion about the y axis. In general, it is necessary to remember that you are adding up cylindrical
shells and to carefully figure out the height and radius, plugging into A D 2 rh to get the for-
mula to integrate. Care is also needed in figuring out the limits of integration. Next, we provide
an example of a problem where we use disks to rotate about the y axis.
In order to do this we need to use inverse functions. These were defined in Fast Start Differential
Calculus. Examples of inverse functions include the following.
p
• For x  0, when f .x/ D x 2 we have f 1 .x/ D x .

• If g.x/ D ex , we have g 1
.x/ D ln.x/.

• If h.x/ D tan.x/, we have h 1 .x/ D tan 1 .x/.

The notation for “inverse” and “negative first power” are identical and can only be told apart by
examining context. Be careful!

Example 3.5 Rotation about the y -axis with disks: Find the volume of rotation when the
area bounded by y D x 2 , the y -axis, and y D 4 is rotated about the y axis.

.2; 4/

x
.0; 0/ .2; 0/
78 3. ADVANCED INTEGRATION
Solution:

For this problem, the radius of the disks is going from the y -axis to the curve y D x 2 .
We need to solve this equation for x which is not too difficult:
p
x D g.y/ D y

We ignore the negative square root because we obviously need a positive radius. This means
that we have a disk radius of x D g.y/.

The volume formula becomes Z y1

V D g.y/2
dy
yo

The function x D g.y/ is the inverse function of y D f .x/. We can now do the calculations for
volume:
Z 4
p
Volume D  . y/2
dy
0

Z 4
D y
dy
0

 2 ˇˇ4
D y ˇ
2 0


D .16 0/
2

D 8 units3

To see that is it possible, let’s set this up (but not calculate the integral) with cylinders.

The height of a cylinder is from y D x 2 to y D 4 meaning that h D 4 x 2 . So, the vol-

ume integral is
Z xD2
V D 2 x.4 x 2 /
dx
xD0

Not a terribly hard integral, but one that is harder than the disk integral.

˙
 3.1. VOLUMES OF ROTATION 79
Knowledge Box 3.3

Disks or Cylinders?
Both methods for finding volume of rotation involve adding up areas
with integration to find a volume. The method of disks adds up circular
disks and the method of cylinders adds up, well, cylinders. How do you
tell which method to use?

You use whichever method you can set up and, if you can set up
both, you use the one that yields the easier integral.

The next example is another one that lets us practice with the method of cylinders. It is an exam-
ple where the calculations to find the radius values needed to use the method of disks is too hard.

Example 3.6 Compute the volume obtained by rotating the area bounded by the curve

y D 2x 2 x3

and the x -axis around the y -axis.

Solution:

x
.0; 0/ .2; 0/

Let’s do the calculations. The points where y D 0 are x D 0 and x D 2, giving us the limits of
integration. In this case the radius of the cylinders is just x , and the height is just y . So, applying
80 3. ADVANCED INTEGRATION
the formula from Knowledge Box 3.2 we get:

Z 2
Volume D 2 x.2x 2 x 3 /
dx
0

Z 2
D 2 .2x 3 x 4 /
dx
0

ˇ !
1 4 1 5 ˇˇ2
D 2 x x
2 5 ˇ0

 
32
D 2 8 0C0
5

8
D 2

5

16
D units3
5

If we wanted to use the method of disks we would need to solve y D 2x 2 x 3 for x to get the
inverse function – a challenging piece of algebra.

Suppose that we want to rotate the area between two curves about the x -axis. Then we get a
large disk for the outer curve and a small disk for the inner curve – meaning that we get washers.
A rendering of a washer is shown in Figure 3.4. The area of a washer with outer radius r1 and
inner radius r2 is the difference of the area of the overall disk and the missing inner disk:

A D  r12  r22 D .r12 r22 /

This area formula forms the basis of the integration performed with the method of washers.
 3.1. VOLUMES OF ROTATION 81

Figure 3.4: A washer or disk with a hole in the center. When we want to rotate the difference of
two curves about the x -axis, this shape replaces the disks used when only one curve is rotated.

Knowledge Box 3.4

Formula for finding volume with the method of washers

If y D f1 .x/ and y D f2 .x/ are functions for which f1 .x/  f2 .x/ on the in-
terval Œa; b, then the volume obtained by rotating the area between the curves
about the x -axis is:
Z b


V D f1 .x/2 f2 .x/2
dx:
a

(Notice that we took the constant  out in front of the integral sign.)
82 3. ADVANCED INTEGRATION
Example 3.7 Find the volume resulting from rotating the area between f1 .x/ D x 2 and
p
f2 .x/ D x about the x -axis.

Solution:

To apply the method of washers, we need to know where the curves intersect and which
one is on the outside edge of the washers. It is easy to see that they intersect at .0; 0/ and .1; 1/.
So the limits of integration will be from x D 0 to x D 1. On the range 0  x  1, it’s easy to
p
see that x 2  x . So, the outer curve is f2 .x/. This gives us enough information to set up the
integral.
y

y2 D x 2
.1; 1/ p
y1 D x

x
.0; 0/

Z 1

Volume D  f1 .x/2 f2 .x/2
dx
0

Z 1

D x x 4 dx
0

 ˇ
1 2 1 5 ˇˇ1
D x x ˇ
2 5 0

3
D units3
10
˙
 3.1. VOLUMES OF ROTATION 83
There are a large number of different ways to rotate objects about the x - and y -axis. While
formulas are presented in this section, it is a really good idea to identify the radius and height
of the disks, cylindrical shells, or washers you are integrating to create a volume of rotation.
Without that understanding, it is easy to pick the wrong formula. For Example 3.4 none of the
formulas apply!
The formulas given in this section demonstrate how to add up infinitely thin disks, cylinders,
and washers with an integral to compute a volume. These are relatively intuitive shapes with
which we should already be familiar. Calculus is not constrained by familiarity: an integral can
add up any sort of shapes. If you want a challenge, try to find the formula for the volume of a
pyramid with a square base. The formula should be based on the side length of the square base
and the height of the pyramid.

PROBLEMS
Problem 3.8 Find the volume of rotation about the x -axis for each of the functions below
from x D 0 to x D 1.

1. f .x/ D 3x 5 4. r.x/ D 12x 5 7. a.x/ D cos.x/

2. g.x/ D 0:5x 6 5. s.x/ D 6x 11
3. h.x/ D 5x 7 6. q.x/ D e x
8. b.x/ D .ex C e x
/

Problem 3.9 If we rotate y D sin.x/ about the x -axis, the result is a string of beads. Find the
volume of one bead.

Problem 3.10 Repeat Problem 3.9 for f .x/ D sin.4x/.

Problem 3.11 Recalculate Example 3.5 using the method of cylinders.

Problem 3.12 Verify the volume formula

4 3
V D r
3
for a sphere using the method of disks about the x -axis.

Problem 3.13 Find the value a so that rotating the curve y D ln.x/ about the x -axis from
x D 0 to x D a yields a volume of 4 units3 .
84 3. ADVANCED INTEGRATION
Problem 3.14 Find the volume of rotation about the y axis of the area below each of the
functions listed from x D 0 to x D 1.

1. f .x/ D 6x 9 4. r.x/ D 5x 8 7. a.x/ D cos.x/

2. g.x/ D 16x 8 5. s.x/ D 11x 8
x
3. h.x/ D 3x 12 6. q.x/ D ex 8. b.x/ D x 2 C1

Problem 3.15 We know the area under y D 1=x on the interval Œ1; 1/ is infinite. Find the
volume of rotation of y D 1=x on this interval.

Problem 3.16 For each of the following functions and starting and ending x values, find the
volume of rotation of the function about the x -axis.

2=5
1. f .x/ D x between x D 3 and x D 7 4. f .x/ D x 3=7 between x D 5 and x D 7
2=5 3
2. f .x/ D x between x D 3 and x D 5 5. f .x/ D x between x D 5 and x D 9
3. f .x/ D x 2=7 between x D 4 and x D 7 6. f .x/ D x 2
between x D 8 and x D 9

Problem 3.17 For each of the following pairs of functions, compute the volume obtained by
rotating the area between the functions about the x -axis.

1. f .x/ D 3x 3 and g.x/ D 9x 2 4. f .x/ D 3x 4 and g.x/ D 3x 3

2. f .x/ D x 3 and g.x/ D 4x 2 5. f .x/ D 2x 3 and g.x/ D 6x 2
3. f .x/ D 3x 5 and g.x/ D 48x 6. f .x/ D x 3 and g.x/ D x 2
 3.2. ARC LENGTH AND SURFACE AREA 85
Problem 3.18 For each of the following pairs of functions, compute the volume obtained by
rotating the area between the functions about the y -axis.

1. f .x/ D 7x 3 and g.x/ D 28x 2 4. f .x/ D 3x 6 and g.x/ D 3x 2

2. f .x/ D 4x 6 and g.x/ D 64x 2 5. f .x/ D 7x 5 and g.x/ D 21x 4
3. f .x/ D 2x 4 and g.x/ D 18x 2 6. f .x/ D 2x 3 and g.x/ D 8x

Problem 3.19 Using the method of disks, rotating about the x -axis, verify the formula for the
volume of a cone of radius R and height H :

1
V D 
R2 H
3

3.2 ARC LENGTH AND SURFACE AREA

In this section we will learn to compute the length of curves and, having done that, to find the
surface area of figures of rotation.

A piece of a curve is called an arc. The key to finding the length of an arc is the differ-
ential of arc length.

In the past we have had quantities like dx and dy that measure infinitesimal changes in
the directions of the variables x and y .

The differential of arc length is different – it does not point in a consistent direction,
rather it points along a curve and so, by integrating it, we can find the length of a curve.

Examine Figure 3.5. The relationship between the change in x and y and the change in the
length of the curve is Pythagorean, based on a right triangle.

If we take this relationship to the infinitesimal scale, we obtain a formula for the differ-
ential of arc length.
86 3. ADVANCED INTEGRATION

4
y=f(x)

s
y

x

-2 4

-1

Figure 3.5: The triangle with sides x , y , and s shows how the change in the length of the
curve is related to the changes in distance in the x and y directions.
 3.2. ARC LENGTH AND SURFACE AREA 87
Knowledge Box 3.5

The differential of arc length and arc length

If y D f .x/ is a continuous curve, then the rate at which the length
of the graph of f .x/ changes is called the differential of arc length,
denoted by ds . The value of ds is:

ds 2 D dy 2 C dx 2
p
ds D dy 2 C dx 2
s 
dy 2
D C 1
dx 2
dx 2
q
D .y 0 /2 C 1
dx:

The length S of a curve (arclength of the curve) from x D a to x D b is:

Z b
SD ds:
a

Example 3.20 Find the length of y D 3x 2=3 from x D 1 to x D 8.

Solution:

The first step in an arc length problem is to compute ds .

y D 3x 2=3

y 0 D 2x 1=3

p
ds D 4x 2=3 C 1
dx
88 3. ADVANCED INTEGRATION
This means that the desired length is
Z 8
SD ds
Z1 8 r
4
D C 1
dx
1 x 2=3
s
Z 8
4 C x 2=3
D
dx
1 x 2=3
Z 8 p dx
D 4 C x 2=3

1 x 1=3
2 1=3 2 dx
Let u D 4 C x 2=3 , then du D x
dx D
3 3 x 1=3
3 dx
So, du D 1=3 .
2 x
Applying the substitution to the limits we see that the integral goes from u D 5 to u D 8.
Transforming everything to u-space, the arc length is:
Z 8
p 3
SD u
du
5 2
Z 8
3
D u1=2 du
2 5

 ˇ
3 2 3=2 ˇˇ8
D u ˇ
2 3 5

D 83=2 53=2

Š 11:45 units2
˙

Alert students will have noticed that the function chosen to demonstrate arc length is not one
of our usual go-to functions for demonstration. This is because the formula for ds yields some
very difficult integrals. The next example is one such, but yields a formula we already know how
to integrate.
 3.2. ARC LENGTH AND SURFACE AREA 89
2
Example 3.21 Find the arc length of y D x from x D 0 to x D 2.

Solution:

p
Since y 0 D 2x , it is easy to find that ds D 4x 2 C 1
dx , meaning our integral is:

Z 2 p
SD 4x 2 C 1
dx
0

This is a trig-substitution integral. The triangle for this integral is

p
4x 2 C 1
2x


1

which means our substitutions are:

p
4x 2 C 1 D sec. /

2x D tan. /

x D .1=2/ tan./

dx D .1=2/ sec2 ./
d

So we get:
90 3. ADVANCED INTEGRATION

Z 2 p Z ‹
4x 2 C 1
dx D sec. /
.1=2/ sec2 . /
d
0 ‹

Z ‹
1
D sec3 ./
d (known integral)
2 ‹

ˇ‹
1 ˇ
D .sec./ tan./ C ln j sec./ C tan. /j/ˇˇ
4 ‹

1 p 2 p ˇˇ2
D 4x C 1
2x C ln j 4x C 1 C 2xj ˇˇ
2
4 0

Now that we have performed the integral and transformed it back into x -space we can substitute
in the limits and get the arc length.

1 p p p p 
SD 17
4 C ln j 17 C 4j 1
0 C ln j 1 C 0j
4
1 p p 
D 4 17 C ln. 17 C 4/ Š 4:65 units
4

Arc-lengths integrals are often challenging. Let’s do one more example with a function chosen
to keep the difficulty from getting out of hand. This example looks for a general formula for the
arc length of a function.

Example 3.22 Find the length of y D x 3=2 from x D 0 to x D a.

Solution:
3
Since y 0 D x 1=2 we see that
2
r
9 1p
ds D x C 1
dx D 9x C 4
dx
4 2
 3.2. ARC LENGTH AND SURFACE AREA 91
This means that the desired arc length is:

Z a
1p
SD 9x C 4
dx
0 2
Z a
1 p
D 9x C 4
dx
2 0

1
Let u D 9x C 4 so that du D dx
9
Z ‹
1 p 1
D u
du
2 ‹ 9
Z ‹
1
D u1=2
du
18 ‹

 ˇ
1 2 3=2 ˇˇ‹
D u ˇ Need to substitute back to x
18 3 ‹

1 ˇa
ˇ
D .9x C 4/3=2 ˇ
27 0

.9a C 4/3=2 8
D
27

which is the desired arc length formula.

˙

We have already computed the volume of a solid that is enclosed by the graph of a function
rotated about the x -axis. The solids defined in this fashion also have a surface area, the slices of
which are circles.

Figure 3.6 shows examples of the circles that appear in such a rotation.
92 3. ADVANCED INTEGRATION
y
(4,2)
y D f .x/

Figure 3.6: Shown are some circles that are the “slices” of the surface area obtained by rotating
f .x/ about the x -axis.

Using Figure 3.6 as inspiration, we can compute surface area by integrating the circumference
of circles with radius f .x/. Since these circle follow the arc of f .x/, the correct type of change
is the differential of arc length, ds . This leads to the formula in Knowledge Box 3.6.

Knowledge Box 3.6

Surface area of rotation

If y D f .x/ is a continuous curve, then the surface area A obtained by
rotating f .x/ around the x -axis from x D a to x D b is:
Z b
A D 2 f .x/
ds:
a
 3.2. ARC LENGTH AND SURFACE AREA 93
p
Example 3.23 Find the surface area of rotation of y D x about the x -axis from x D 0 to
x D 4.

Solution:
r
1=2 01 1=2 1 1
Since y D x , y D x D p . So: ds D C 1
dx Using the formula from
2 2 x 4x
Knowledge Box 3.6 we obtain the surface area integral,
Z 4 r
p 1
A D 2 x
C 1
dx
0 4x

Z s  
4
1
D 2 x
C 1 dP x
0 4x

Z 4
r
1 C 4x
D 2
dx
0 4
Z 4
1p
D 2 1 C 4x
dx
0 2
Z 4 p
D 1 C 4x
dx
0

1
Let u D 4x C 1, du D dx
4
Z 17
1
D u1=2
du
1 4
Z 17

D u1=2
du
4 1

 2 3=2 ˇˇ17
D u ˇ
43 1

  3=2 
D 17 1 Š 36:18 units2
6
˙
94 3. ADVANCED INTEGRATION
Example 3.24 Find the surface area of rotation for y D ex from x D 0 to x D 2.

Solution:

Noting that y 0 D ex , we see that

p
ds D e2x C 1
dx:

This means the surface area is:

Z 2 p
A D 2 ex e2x C 1
dx
0

Z 2 p
D 2 e2x C 1 .ex
dx/
0

Let u D ex , then du D ex
dx
Z ‹ p
D 2 u2 C 1
du (known integral)
‹

 p p ˇ‹
ˇ
D 2 u u2 C 1 C ln ju C u2 C 1j ˇ
‹

 p p ˇ2
ˇ
D 2 ex e2x C 1 C ln jex C e2x C 1j ˇ
0

 p p p p 
D 2 e2 e4 C 1 C ln.e2 C e4 C 1/ 2 ln j1 C 2j units2

The types of integrals that arise from arc length and rotational surface area problems are often
quite challenging. This justifies the large number of integration techniques we learned when we
studied methods of integration, several of which came up in this section.
If you study multivariate calculus more deeply, the differential of arc length will appear again for
tasks like computing the work done moving a particle along a path through a field. This section
is a bare introduction to the power and applications of the differential of arc length.
 3.2. ARC LENGTH AND SURFACE AREA 95
PROBLEMS

Problem 3.25 For each of the following functions, compute ds , the differential of arc length.

x
1. f .x/ D x 3 5. s.x/ D e

2. g.x/ D sin.x/ 6. q.x/ D x 3=4

3. h.x/ D tan 1 .x/ 1

7. a.x/ D
x 3
4. r.x/ D 8. b.x/ D 2x
xC1

Problem 3.26 For each of the following functions, compute the arc length of the graph of the
function on the given interval.

p
1. f .x/ D 9x 2=3 I Œ0; 1 5. s.x/ D .x 2/3 I Œ4; 5

2. g.x/ D 2x C 1I Œ0; 4 6. q.x/ D .x C 1/2=3 I Œ 1; 1

3. h.x/ D 2x 3=2 I Œ2; 5 7. a.x/ D 4x 3=2 I Œ2; 4
4. r.x/ D x 2 C 4x C 4I Œ0; 6 8. b.x/ D x 2 C x C 1=4I Œ0; 8

Problem 3.27 For each of the following functions, compute the surface area of rotation of the
function for the given interval.

1. f .x/ D xI Œ0; 3 4. r.x/ D sin.x/I Œ0; 

p
2. g.x/ D xI Œ4; 9 5. s.x/ D cos.3x/I Œ0; =4
3. h.x/ D ex I Œ0; 1 6. q.x/ D sin.x/ cos.x/I Œ0; =2
96 3. ADVANCED INTEGRATION

Problem 3.28 Using the techniques for surface area of revolution, find the formula for the
surface area of a cone with apex angle  , as shown above. Don’t forget the area of the bottom.

Problem 3.29 If
y D x 2=3 ;
find the formula for the arc length of the graph of this function on the interval Œa; b .

Problem 3.30 Based on the material in this section, if

.f .t /; g.t //

is a parametric curve, what would the differential of arc length, ds , be?

Problem 3.31 Derive the polar differential of arc length.

Problem 3.32 Derive the parametric differential of arc length for

.x.t /; y.t //

Problem 3.33 Find but do not evaluate the integral for computing the arc length of

y D sin.x/:

Discuss: what techniques might work for this integral.

 3.3. MULTIPLE INTEGRALS 97
Problem 3.34 Which is harder, finding the arc-length of
p
yD x

or its surface area of rotation? Why?

Problem 3.35 We already know the area under y D 1=x on the interval Œ1; 1/ is infinite but
that the enclosed volume of rotation is finite. Using comparison and cleverness, demonstrate the
surface area of this shape is infinite.

3.3 MULTIPLE INTEGRALS

In our earlier study of integration, we learned to use integrals to find the area under a curve. The
analogous task in three dimensions is to find the volume under a surface over some domain in
the x -y plane. As with partial derivatives we will find the idea of a currently active variable useful.

When we were integrating a single-variable function to obtain an area, we integrated over an

interval on the x -axis. When we are finding volumes under surfaces, we will integrate over regions
or subsets of the plane. An example of a relatively simple region in the plane is shown in Figure
3.7.

(2,2)

-1 3

-1

Figure 3.7: A rectangular region R, 0  x; y  2 in the plane.

98 3. ADVANCED INTEGRATION
So far in this chapter we have found volumes by using integrals to add up slices. To perform
volume integrals we will have to integrate slices that are, themselves, the result of integrating.
This means we will use multiple or iterated integrals. Fortunately, these are done one at a
time. We use differentials, like dy and dx , to cue which integral is performed next. Like forks
at a fancy dinner party, the next integral is the one for the differential farthest to the left.

Example 3.36 Find the volume under

f .x; y/ D x 2 C y 2

that is above R, 0  x; y  2, as shown in Figure 3.7.

Solution:

The y and x variables are both in intervals of the form Œ0; 2, so the limits of integration
are 0 and 2 for both the integrals we must perform. The volume under the surface is:

Z 2 Z 2

V D x 2 C y 2
dx dy The variable x is active.
0 0

Z 2  ˇ2
1 3 ˇ
D x C xy 2 ˇˇ
dy
0 3 0

Z 2  
8
D C 2y 2 0 0
dy
0 3
Z 2  
8
D 2y 2 C
dy Now y is active.
0 3
ˇ
2 3 8 ˇˇ2
D y C yˇ
3 3 0
 
16 16 32
D C 0 0 D units3
3 3 3

Example 3.36 contained two integrals. During the first, x was the active variable, and y acted
like a constant; when the limits of integration were substituted in they were substituted in for x
 3.3. MULTIPLE INTEGRALS 99
not y . The second integral took place in an environment where x was gone and y was the active
variable.

Knowledge Box 3.7

Integration with respect to a variable

When we are computing an integral
Z bZ d
V D f .x; y/
dx dy;
a c

the first integral treats x as a variable and is said to be an integral with

respect to x ; the second integral treats y as a variable and is said to be
an integral with respect to y .

When we are integrating with respect to one variable, other variables act as constants and so can
pass through integral signs. This permits us to simplify some integrals.

Example 3.37 Use the constant status of variables to compute

Z 1 Z 1

x 2 y 2
dx dy
0 0

in a simple way.

Solution:

Z 1 Z 1 Z 1 Z 1 
2 2

2 2
x y
dx dy D y
dy x
dx Pass y through the x integral
0 0 0 0

Z 1  Z 1 
2 2
D y
dy
x
dx
0 0

Z 1 2
2
D x
dx Since the integrals are equal
0
100 3. ADVANCED INTEGRATION
ˇ !2
1 3 ˇˇ1
D x
3 ˇ0

 2
1 1
D 0 D units3
3 9
˙

This sort of integral – that can be split up into two different integrals – is called a decomposable
integral.
Knowledge Box 3.8

Decomposable Integrals
The multiple integral of the product of a function of one variable by a
function of the other variable can be factored into two single-variable
integrals.
Z Z Z  Z 
f .x/g.y/
dx dy D f .x/ dx g.y/ dy :

Example 3.38 Use the decomposition of integrals to perform the following:

Z =2 Z 2
.x
cos.y//
dx dy
0 0

Solution:

Z Z Z  Z !
=2 2 2 =2
.x
cos.y//
dx dy D x dx  cos.y/ dy
0 0 0 0

ˇ2 ! ˇ=2 !
x 2 ˇˇ ˇ
ˇ
D  sin.y/
2 ˇ0 ˇ
0

D .4=2 0=2/  .sin.=2/ sin.0// D 2
1 D 2

˙
 3.3. MULTIPLE INTEGRALS 101
Volume integration becomes more difficult when the region R is not a rectangle. Over a
rectangular region, the limits of integration are constants. If a region is not rectangular, then
the curves that describe the boundaries of the region become involved in the limits of integration.

Example 3.39 Integrate the function f .x; y/ D 2x y C 4 over the region R given in Figure
3.8.

Solution:

The function is simple, but the limits of integration are tricky. In this case, 0  x  2
and, for a given value of x , 0  y  x . This is because the upper edge of the region of integration
is the line y D x .

Now, we can set up the integral, choosing the order of integration to agree with the limits.
Z 2 Z x
D .2x y C 4/ dy dx
0 0

Z 2  ˇx
1 2 ˇ
D 2xy y C 4y ˇˇ dx
0 2 0

Z 2  
1 2
D 2x 2 x C 4x 0 0 0
dx
0 2
Z 2  
3 2
D x C 4x dx
0 2
ˇ2
1 3 ˇ
2ˇ
D x C 2x ˇ
2 0

1
D 8C8 0 0
2

D 12 units3

˙
102 3. ADVANCED INTEGRATION
3

(2,2)

R
-1 3

-1

Figure 3.8: A non-rectangular region bounded by the x -axis, the line x D 2, and the line y D x .

Example 3.40 Find the volume underneath

f .x; y/ D 4 x2

over the region bounded by

yDx

y D 2x

and
xD2

Solution:

Start by drawing the region of integration.

This region has x bounds 0  x  2.

For a given value of x the region goes from the line y D x to the line y D 2x .
 3.3. MULTIPLE INTEGRALS 103

-1 3

-1

This gives us sufficient information to set up the integral.

Z 2 Z 2x

V D 4 x 2 dy dx
0 x

Z ˇ
2
ˇ2x
D 4y x y ˇˇ
dx
2
0 x

Z 2

D 8x 2x 3 4x C x 3 dx
0

Z 2

D 4x x 3 dx
0

ˇ
2 1 4 ˇˇ2
D 2x x
4 ˇ0

1
D8
16 0C0
4

D 4 units3

˙
104 3. ADVANCED INTEGRATION
The regions we have used thus far have been based on functions that are easy to work with using
Cartesian coordinates.

To deal with other sorts of regions, we need to first develop a broader point of view.

R
-3 3

-3

Figure 3.9: Another sort of region – a disk of radius 2 centered at the origin.
 3.3. MULTIPLE INTEGRALS 105
Knowledge Box 3.9

The differential of area

The change of area is
dA D dx dy D dy dx:
This permits us to change our integral notation to the following for the
integral of f .x; y/ over a region R
Z Z
V D f .x; y/ dA:
R

In polar coordinates:
dA D r dr d:

Example 3.41 Find the area enclosed below the curve f .x; y/ D x 2 C y 2 over a disk of radius
2 centered at the origin.

Solution:

This volume is below the function f .x; y/ for points no farther from the origin than 2.
This means the region R is

x 2 C y 2  4:

In polar coordinates, this region is those points .r;  / for which 0  r  2 and 0    2 .
Since the polar/rectangular conversion equations tell us r 2 D x 2 C y 2 , f .r;  / D r 2 .

Using polar coordinates the integral is:

Z Z
V D f .r;  / dA
R

Z 2 Z 2
D r 2
r
dr d
0 0
106 3. ADVANCED INTEGRATION
Z 2 Z 2
D r 3
dr d
0 0

Z ˇ
2
1 4 ˇˇ2
D r d
0 4 ˇ0
Z 2
D .16=4 0/ d
0

Z 2
D 4d
0

ˇ2
ˇ
D 4 ˇˇ
0

D 8 0 D 8 units3

The next example is a very important one for the theory of statistics. As you know if you have
studied statistics, the normal distribution has a probability distribution function of:

1 x 2 =2
p e
2

The area under the curve of a probability distribution function must be equal to one. Thus,
if you have a function with an area greater than one, you must multiply it by a normalizing
constant equal to one over the area.
1
The following example shows where the normalizing constant p in the normal distri-
2
bution probability distribution function comes from.
The integral relies on a trick: squaring the integral and then shifting the squared integral to polar
coordinates. This changes an impossible integral into one that can be done without difficulty by
u-substitution. Sadly, this only permits the evaluation of the integral on the interval Œ 1; 1;
the coordinate change is intractable except on the full interval where the function exists.
 3.3. MULTIPLE INTEGRALS 107
Z 1
x 2 =2
Example 3.42 Find A D e
dx :
1

Solution:

The solution works if you compute the square of the integral.

Z 1 2
x 2 =2
2
A D e
dx
1

Z 1  Z 1 
x 2 =2 x 2 =2
D e
dx e
dx
1 1

Z 1  Z 1 
x 2 =2 y 2 =2
D e
dx e
dy Rename
1 1

Z 1 Z 1
x 2 =2 y 2 =2
D e e
dy dx
1 1

Z 1 Z 1
1=2.x 2 Cy 2 /
D e
dA
1 1

Change to polar coordinates

Z Z
1=2.r 2 /
D e
dA
R

Z Z
1=2.r 2 /
D e
r
dr d
R

The polar region in question is 0  r < 1 and 0   < 2 . Rebuild the integral with these
limits and we get:
Z 2 Z 1
2
A2 D r
e r =2
dr d
0 0

Z 2  Z 1 
r 2 =2
D d
r
e
dr
0 0

ˇ2 Z 1 
ˇ r 2 =2
D  ˇˇ
r
e
dr
0 0
108 3. ADVANCED INTEGRATION
Z 1
r 2 =2
D 2 r
e
dr
0

Z a
r 2 =2
D 2 lim r
e
dr
a!1 0

Let u D r 2 =2, then du D r
dr

Z ‹
D 2 lim
a!1 ‹
eu
du

Z ‹
D 2 lim
a!1 ‹
eu
du
ˇa
r 2 =2 ˇ
D 2 lim
a!1
e ˇ
0

 
a2 =2
D 2 lim
a!1
e 1

D 2.0 1/ D 2

p
If A2 D 2 then A D 2 , which is the correct normalizing constant.

3.3.1 MASS AND CENTER OF MASS

The center of mass for an object is the average position of all the mass in an object. This section
demonstrates techniques for computing the center of mass of flat plates with a density function
.x; y/. Density is the rate at which mass changes as you move through an object, which, in
turn, means that the mass of an object is the integral of its density.
 3.3. MULTIPLE INTEGRALS 109
Knowledge Box 3.10

Mass of a plate
Suppose that a flat plate occupies a region R with a density function
.x; y/ defined on R. Then the mass of the plate is
Z Z
M D .x; y/
dA:
R

Remember that the function .x; y/ is usually constant, or close enough to constant that we
assume it to be constant, when we have a mass made of a relatively uniform material. The fairly
high variation in the mass functions in the examples and homework problems is intended to give
your integration skills a workout – not as a representation of situations encountered in physical
reality.
Example 3.43 If a plate fills the triangular region R from Figure 3.8 with a density function
.x; y/ D x C 1 grams/unit2 , find the mass of the plate.

Solution:

Using the mass formula, the integral is

Z 2 Z x
Mass D .x C 1/ dy dx
0 0

Z 2
ˇx
ˇ
D .xy C y/ ˇˇ
dx
0 0

Z 2

D x 2 C x
dx
0

ˇ2
x3 x 2 ˇˇ
D C
3 2 ˇ0

D 8=3 C 2 0 0 D 14=3 g

˙
110 3. ADVANCED INTEGRATION
Once we have the ability to compute the mass of a plate from its dimensions and density, we
can compute the coordinates of the center of mass of the plate using a type of averaging integral.

Knowledge Box 3.11

Center of mass
Suppose that a flat plate occupies a region R with a density function
.x; y/ defined on R. Then if
Z Z
Mx D y.x; y/
dA
R

and Z Z
My D x.x; y/
dA
R
the center of mass of the plate is
 
My Mx
.x; y/ D ; :
M M

-2 2

-2

Figure 3.10: Center of mass of square region when .x; y/ D 2y g/unit2 .

 3.3. MULTIPLE INTEGRALS 111
Example 3.44 Suppose that R is the square region

0  x; y  1

and that .x; y/ D 2y g/unit2 as shown in Figure 3.10.

Find the center of mass.

Solution:

This problem requires three integrals.

Z 1 Z 1 Z 1 Z 1 Z 1 Z 1
M D 2y
dy dx Mx D y
2y
dy dx My D x
2y
dy dx
0 0 0 0 0 0

Z 1
ˇ1 Z 1 Z 1 Z 1
ˇ1
ˇ
2ˇ 2
ˇ 2ˇ
D y ˇ dx D 2y
dy dx D xy ˇ
dx
0 0 0 0 0 0

Z Z ˇ Z
1 1
2 3 ˇˇ1 1
D .1 0/dx D y ˇ
dx D .x 0/
dx
0 0 3 0 0

Z Z   ˇ
1 1
2 1 2 ˇˇ1 1
D dx D 0
dx D x ˇ D
0 0 3 2 0 2
ˇ1 Z 1
ˇ 2
D x ˇˇ D
dx
0 0 3
ˇ1
2 ˇ 2
D .1 0/ D 1 gram D x ˇˇ D
3 3 0

Now that we have the pieces we can use the formula for center of mass:
   
1=2 2=3 1 2
.x; y/ D ; D ;
1 1 2 3

˙
112 3. ADVANCED INTEGRATION
PROBLEMS

Problem 3.45 Find the integral of each of the following functions over the specified region.
1. The function f .x; y/ D x C y 2 on the strip

0x4 0  y  1:

2. The function g.x; y/ D xy on the rectangle

1x3 1  y  2:

3. The function h.x; y/ D x 2 y C xy 2 on the square

0x2 0  y  2:

4. The function r.x; y/ D 2x C 3y C 1 on the region bounded by x D 0, y D 1, and y D x .

5. The function s.x; y/ D x 2 C y 2 C 1 on the region bounded by the x axis and the function
y D 4 x2.
p
6. The function q.x; y/ D x C y on the region bounded by the curves y D x and y D x 2 .

7. The function a.x; y/ D x 2 on the region bounded by the curves y D 2x and y D x 2 .

p
8. The function b.x; y/ D y 2 on the region bounded by the curves y D 3 x and y D x for
x  0.

Problem 3.46 Sketch the regions from Problem 3.45.

Problem 3.47 Explain why a density function .x; y/ can never be negative.

Problem 3.48 Find a region R so that the integral over R of f .x/ D x 2 C y 2 is 6 units3 .

Problem 3.49 Find the square region 0  x; y  a so that

Z Z


x 3 C y
dA
R

is 12 units3 .
 3.3. MULTIPLE INTEGRALS 113
Problem 3.50 Find the mass of the plate 0  x; y  3 if

.x; y/ D y 2 C 1:

Problem 3.51 Find the center of mass of the region bounded by the x -axis, the y -axis, and
the line x C y D 4 if the density function is

.x; y/ D y C 1:

p
Problem 3.52 Find the center of mass of the region bounded by y D x and y D x 2 if the
density function is
.x; y/ D x C 2:

Problem 3.53 Find the center of mass of the region 0  x; y  1 if the density function is

.x; y/ D .x C y/=2:

Problem 3.54 Find the volume under

p
f .x; y/ D x2 C y2

above the region x 2 C y 2  16.

Problem 3.55 Find the volume under

f .x; y/ D x 2 C y 2

above the region bounded by the petal curve r D 2 cos.3 /.

Problem 3.56 Find the volume under

f .x; y/ D .x 2 C y 2 /3=2

above the region bounded by the petal curve r D cos.2/.

114 3. ADVANCED INTEGRATION
Problem 3.57 Find a plane f .x; y/ so that the area under the plane but over a circle of radius
2 centered at the origin is 16 units3 .

Problem 3.58 Derive the general formula for the volume over a rectangle and under a plane
in a region where the plane has a positive z value.
 115

CHAPTER 4

Sequences, Series, and

Function Approximation
This chapter is quite different from the rest of the material in the Fast Start series. Calculus
is mostly part of the mathematics of continuous functions, while the sequences and series we
study in this chapter are a part of discrete mathematics – math that is broken up into individual
pieces. Discrete math is about things you can count, rather than things you can measure. Since
this is a calculus book, we will apply what we learn to understanding and increasing the power of
our calculus. The ultimate goal of the chapter is a much deeper understanding of transcendental
(non-polynomial) functions like ex or cos.x/. We start, however, at the beginning.

4.1 SEQUENCES AND THE GEOMETRIC SERIES

A sequence is an infinite list of numbers. Sometimes we give a sequence by listing an obvious
pattern:
1 1 1
S D 1; ; ; ; : : :
2 3 4
with the ellipsis meaning “and so on.” We also can use more formal set notation to specify a
sequence with a formula:
 
1
SD W n D 1; 2; : : :
n

Like a function having a limit at infinity, there is a notion of a sequence converging.

Knowledge Box 4.1

Definition of the convergence of a sequence to a limit

We call L the limit of a sequence fxn W n D 1; 2; : : :g if, for each  > 0,
there is a whole number N so that, whenever n > N , we have:
jxn Lj < :
A sequence that has a limit is said to converge.
116 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Example 4.1 Prove that the sequence
 
1
SD W n D 1; 2; : : :
n

converges to zero.

Solution:
1 1
For  > 0 choose N to be the smallest whole number greater than . This makes <  . Then
 N
if n > N we have:

n>N

1 1
< Reciprocals reverse inequalities
n N

1
< Since 1=N < 
n
ˇ ˇ
ˇ1 ˇ
ˇ 0ˇˇ <  Value on the left did not change
ˇn

Which satisfies the definition of the limit of the sequence being L D 0.

As always there are shortcuts that mean we only need to rely on the definition of the limits of
sequences occasionally.

Knowledge Box 4.2

Sequences drawn from functions

Suppose that we have a sequence
S D fxn g D ff .n/ W n D 1; 2; : : :g

and that lim f .x/ D L. Then we may conclude that lim xn D L:

x!1 n!1
 4.1. SEQUENCES AND THE GEOMETRIC SERIES 117
Example 4.2 Suppose that we have a sequence
 
1
xn D W n D 0; 1; 2; : : :
1 C n2

Find lim xn .
n!1

Solution:
1
We already know lim D 0. So, using Knowledge Box 4.2, we have that
x!1 1 C x2
lim xn D 0
n!1

The rule in Knowledge Box 4.2 is not reversible.

Example 4.3 Determine if the sequence

S D fcos.2 n/ W n D 0; 1; 2; : : :g

converges.

Solution:

Since lim cos.2x/ jumps back and forth in the range 1  y  1, the function that
x!1
we drew the sequence from does not have a limit.

Leaving that aside, examine a listing of the first several terms of the sequence:

f1; 1; 1; 1; 1; 1; : : :g

This sequence obviously converges to L D 1.

Again: if the function has a limit, then so does the sequence. The reverse need not be
true.

˙
The next sequence resolution technique requires that we define several terms.
118 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Definition 4.1 If a sequence fxn g has the property that xn  xnC1 , then we say that the sequence is
monotone increasing.

Definition 4.2 If a sequence fxn g has the property that xn  xnC1 , then we say that the sequence is
monotone decreasing.

Definition 4.3 If a sequence fxn g is either monotone increasing or monotone decreasing, then we say
the sequence is monotone.

Definition 4.4 If a sequence fxn g has the property that

xn  C

for all n and for some constant C , then we say the sequence is bounded above.

Definition 4.5 If a sequence fxn g has the property that

xn  C

for all n and for some constant C , then we say the sequence is bounded below.

Definition 4.6 If a sequence fxn g is both bounded above and bounded below, then we say the sequence
is bounded.

Knowledge Box 4.3

Bounded monotone sequences

The following types of sequences all converge.
• A monotone increasing sequence that is bounded above.
• A monotone decreasing sequence that is bounded below.
• Any monotone bounded sequence.

The reason that the sequences listed in Knowledge Box 4.3 converge is that they are required
to move toward a bound of some sort without passing it. This is another useful shortcut for
 4.1. SEQUENCES AND THE GEOMETRIC SERIES 119
determining if a sequence converges, although this shortcut does not tell us the value of the
limit.

Example 4.4 Demonstrate that the sequence

 
n
xn D W n D 0; 1; 2; 3; : : :
nC1

has a limit.

Solution:

While we could use Knowledge Box 4.2, let’s use this example as a chance to demon-
strate the technique from Knowledge Box 4.3.

First notice that this sequence is monotone increasing. To see this, notice that:

n2 C 2n < n2 C 2n C 1

n.n C 2/ < .n C 1/2

n nC1
<
nC1 nC2

xn < xnC1

n
Since n < n C 1, we can deduce that xn D < 1.
nC1
So the sequence is bounded above. This permits us to use Knowledge Box 4.3 to deduce
that the sequence has a limit.

It would have been easier to do Example 4.4 with Knowledge Box 4.2. But there will be times
when the sequence is not drawn from a function when we have to use monotone sequence theory.
The next Knowledge Box extends the reach of our ability to check sequences for convergence.
120 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Knowledge Box 4.4

Arithmetic combinations of sequences

If fxn g has a limit of L, and fyn g has a limit of M , and if a; b are
constants, then:
• lim axn ˙ byn D aL ˙ bM
n!1

• lim xn
yn D L
M
n!1

xn L
• lim D if M ¤ 0.
n!1 yn M
• lim xnk D Lk
n!1

Example 4.5 Find the limit of

 
1 n
S D xn D C 3
W n D 1; 2; 3; : : :
n nC1

Solution:

We already know the limit of 1=n, as a series, is zero. Using Knowledge Box 4.2 it is
easy to see that:
n
lim D1
n!1 n C 1

Combining these results using the information in Knowledge Box 4.4 we get that the limit is:

LD0C3
1D3

˙
This concludes our direct investigation of sequences. We now turn to using sequences as a tool
to explore series. Where a sequence is an infinite list of numbers, a series is an infinite list
of numbers that you add up. This may or may not result in a finite sum – and resolving that
question requires sequence theory.
 4.1. SEQUENCES AND THE GEOMETRIC SERIES 121
Knowledge Box 4.5

Definition of series
If fxn W n D 0; 1; : : :g is a sequence, then
1
X
xn D x0 C x1 C


C xk C



nD0

is the corresponding infinite series. If we sum only finitely many terms

we have a finite series.

Example 4.6 Show that the following infinite series sums to 1.

X1
1n 1 1 1
D C C C



nD1
2 2 4 8

Solution:

1
16 1
8
etc.
1
1
2

1
4

1
Figure 4.1: A unit square cut into pieces 1/2, 1/4, 1/8, : : :

Examine Figure 4.1. The figure divides a unit square (with area 1) into rectangles of size 1/2,
1/4, 1/8, and every other number in the series we are trying to sum. This constitutes a geometric
demonstration that the series sums to 1.

˙
122 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Clearly, finding a cool picture is not a general technique for demonstrating that a series has a
sum. Proving no such picture exists when the series fails to have a sum is even more impossible
– we need a more general theory.

Knowledge Box 4.6

The sequence of partial sums of a series

1
X
Suppose that S D xn is a series. If we set
nD1

n
X
pn D xk ;
kD1

then fpn g is called the sequence of partial sums of S . We say that a

series converges to a sum L if and only if its sequence of partial sums
has L as a limit.

Example 4.7 Find the sequence of partial sums of the series in Example 4.6 and compute its
limit.

Solution:

Compute the first few members of the sequence of partial sums:

p1 D 1=2

p2 D 1=2 C 1=4 D 3=4

p3 D 1=2 C 1=4 C 1=8 D 7=8

p4 D 1=2 C 1=4 C 1=8 C 1=16 D 15=16

Which is a clear pattern, and it is easy to see

1
pn D 1
2n
 4.1. SEQUENCES AND THE GEOMETRIC SERIES 123
This gives us the sequence of partial sums. Computing the limit we get:

1
lim pn D lim 1 D1 0D1
n!1 n!1 2n

So we get the same sum using this more formal approach.

˙
At this point we need to call forward an identity from Fast Start Integral Calculus:

x nC1 1 1 x nC1
1 C x C x2 C


xn D D
x 1 1 x

This identity is one that is true for polynomials, but it also applies to summing a finite series.
Additionally, this formula can be used as the partial sum of a particular type of infinite series –
at least when its limit exists.

Knowledge Box 4.7

Finite and infinite geometric series
The polynomial identity in the text tells us, for a constant a ¤ 1, that:
n
X anC1 1
ak D :
a 1
kD0

This is the finite geometric series formula.

If jaj < 1, then the limit of the finite series gives us the infinite geo-
metric series formula:
1
X 1
an D :
nD0
1 a

Applying Knowledge Box 4.4 we also get that:

1
X c
c
an D
nD0
1 a

for a constant c . The number a is called the ratio of the geometric series.
124 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Example 4.8 Compute
X1
3
5 n
nD0

Solution:

This is a geometric series with ratio a D 1=5. It has a leading constant c D 3. Applying
the appropriate formula we see that:

X1
3 3 3 15
n
D D D
nD0
5 1 1=5 4=5 4

PROBLEMS
Problem 4.9 Prove formally, using the definition of the limit of a sequence, that

fcos.2 n/ W n D 0; 1; 2; : : :g

converges to 1.

Problem 4.10 Prove formally, using the definition of the limit of a sequence, that
 
1
W n D 1; 2; 3; : : :
n2
converges to 0.

Problem 4.11 Prove formally, using the definition of the limit of a sequence, that
 
n
W n D 0; 1; 2; : : :
nC1
converges to 1.
 4.1. SEQUENCES AND THE GEOMETRIC SERIES 125
Problem 4.12 Compute the limit of each of the following sequences or give a reason why the
limit does not exist. Assume n D 1; 2; : : :

˚    
1. xn D tan 1 .n/ n2 3n2
3. zn D 5. zn D 2
nC1 n C1
 
n   o cos.n/
2. yn D sin n 6. zn D
2 4. fyn D sin . n/g nC1

Problem 4.13 Do the calculation to prove the infinite geometric series formula from the finite
one: see Knowledge Box 4.7.

Problem 4.14 Compute the following sums or give a reason they fail to exist.

X1   1
X
20
X 1 n
1. 1:2 k 4. 7. 112 .0:065/n
nD0
4 nD0
kD0
1  n X1  n 1
X 1 3 X
2. 5. 3 8. 2
. 1/n
3 nD0
2
nD0 nD0

X1  n 1
X
1
3. 2
6. 0:05n
nD0
7 nD0

Problem 4.15 Compute the following sums or give a reason they fail to exist.

24
X 30
X 100
X
1. 3n 3. 1:2n 5. 1:1n
kD12 kD3 kD90

15
X 20
X 22
X
2. 2n 4. 4:5n 6. 7n
kD5 kD10 kD1

Problem 4.16 A swinging pendulum is losing energy. Its first swing is 2 m long, and each
after that is 0.9985 times as long as the one before it. Estimate the total distance traveled by the
pendulum.
126 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Problem 4.17 A ball is dropped from a height of 8 m. If each bounce is 3/4 the height of the
one before it, estimate the total vertical distance traveled by the ball.

Problem 4.18 A rod is initially displaced 2.1 mm from equilibrium and undergoes damped
vibration with a decay in the length of each subsequent swing of 0.937. Find the total vertical
distance traveled by the end of the rod.

Problem 4.19 A ball is dropped from a height of 4m. If each bounce is 0.86 times the height
of the one before it, estimate the total vertical distance traveled by the ball.

Problem 4.20 A rod is initially displaced 3cm from equilibrium and undergoes damped vi-
bration with a decay in the length of each subsequent swing of 0.987. Find the total vertical
distance traveled by the end of the rod.

Problem 4.21 A very orderly and goes first north then east over an over. If the distances it
travels before turning go 1, 1/3, 1/9, 1/27, and so on, what the is distance from its starting point
that it approaches as it travels farther and farther?

4.2 SERIES CONVERGENCE TESTS

In this section we will develop a number of tests to determine if a series converges. At present,
we know that an infinite geometric series with a ratio of a with jaj < 1 will converge, but not
much else. We begin with a motivating example.

2 1
yD x

Figure 4.2: Shown is a portion of the graph of y D 1=x and a sequence of rectangles of width 1
and height 1=n for n D 1; 2; 3; : : :.
 4.2. SERIES CONVERGENCE TESTS 127
1
X 1
Example 4.22 Determine if the series has a finite or infinite sum.
nD1
n

Solution:

Examine Figure 4.2.Z 1 This shows that a series of rectangles with area 1; 1=2; 1=3; : : : have
dx
a larger area than because the area under the curve is strictly smaller than the sum of
1 x
the areas of the rectangles. Computing the integral:

Z 1 Z a
dx dx
D lim
1 x a!1 1 x
ˇa
ˇ
D lim ln.x/ˇˇ
a!1
1

D lim ln.a/ ln.1/

a!1

D1

X1
1
shows us that:  1; and we conclude the sum is infinite.
nD1
n
˙

The series
X1
1
nD1
n
is sufficiently important that it has its own name: the harmonic series.

Definition 4.7 When considering

1
X
xn
nD1

we call xn the general term of the series.

Example 4.22 is another example of proving something about a series sum by drawing a clever
picture. An interesting technique, but, as before, we need more general tools. One such tool
involves taking the limit of the sequence generating the series. If it does not approach zero,
128 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
there is no hope that the infinite sum converges.

Knowledge Box 4.8

The divergence test

If lim xn ¤ 0 then
n!1
1
X
xn
nD0

does not have a finite value.

It is important to note that the divergence test is uni-directional. If the general term of a series
does go to zero, that tells you exactly nothing about the behavior of the sum of the series.

Example 4.23 Show that

1
X n
nD0
nC1

diverges (has an infinite sum).

Solution:

Since
n
lim D 1;
n!1 n C 1

the series in question diverges by the divergence test. Colloquially, we are adding up an infinite
number of terms that are approaching one – so the resulting sum is infinite.

When you can use it, the divergence test is often short and sweet. The next test is the formal
version of the test we used in Example 4.22.
 4.2. SERIES CONVERGENCE TESTS 129
Knowledge Box 4.9

The integral test

Suppose that f .x/ is a positive, decreasing function on Π0; 1/ and that
xn D f .n/. Then,
1
X Z 1
xn and f .x/dx
nD1 a

both converge or both diverge for any finite a  1.

Example 4.24 Show that

X1
1
n2
nD1

is finite.

Solution:

Use the integral test.

Z 1 Z a
dx dx
D lim
1 x2 a!1 1 x2
ˇ
1 ˇˇa
D lim
a!1 x ˇ1

1
D lim . 1/
a!1 a

D0C1D1

Since the improper integral is finite, so is the sum. Note that this does not tell us the value of the
sum.
˙

X1 X1
1 1
The fact that the series diverges but 2
converges motivates our next test.
nD1
n nD1
n
130 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Definition 4.8 A p -series is a series of the form

X1
1
nD1
np

where p is a constant.
Knowledge Box 4.10

The p -series test

The p -series
X1
1
nD1
np
converges if p > 1 and diverges if p  1.

The series convergence test in Knowledge Box 4.10 follows directly from the integral test –
something you are asked to verify in the homework.

Example 4.25 Determine if

X1
1
nD1
n

converges to a finite number or diverges.

Solution:

This series has the form of a p -series with p D  . Since  > 1, we conclude the series
converges.

Example 4.26 Determine if

X1
1
p
nD1
n
converges to a finite number or diverges.
 4.2. SERIES CONVERGENCE TESTS 131
Solution:

This series has the form of a p -series with p D 1=2. Since 1=2  1, we conclude the se-
ries diverges.

˙
The next two tests leverage series we already understand to resolve even more series. The tests
depend on various forms of comparison of a series under test to a series with known convergence
or divergence behavior.

Knowledge Box 4.11

The comparison tests

1
X
• If xn converges and 0  yn  xn for all n, we have that
nD0
1
X
yn also converges.
nD0

1
X 1
X
• If xn diverges and yn  xn  0 for all n, we have that yn
nD0 nD0
also diverges.

The colloquial versions of the comparison tests are pretty easy to believe. The first says, of a series
with positive general terms, that if it is smaller, term by term, than another series with a finite
sum, then it has a finite sum. The second reverses that: saying that, if a series is larger, term by
term, than another series with an infinite sum, then it has an infinite sum.

Example 4.27 Determine if

X1
1
nD1
n=3

converges or diverges.
132 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Solution:

Notice that, for n  1,

n  n=3

1 1

n n=3
1 1
So 0   , which permits us to deduce that the series in the example diverges by
n n=3
comparison to the harmonic series, which is known to diverge.

Example 4.28 Determine if

1
X 1
2nC5
nD0

converges or diverges.

Solution:

The key to using the comparison test is to find a good known series to compare to. In
this case the geometric series with ratio 1/2 is natural.

2n  2n C 5

1 1
 n
2n 2 C5
 n
1 1
 n
2 2 C5
Which means that the general term of the convergent geometric series
X1  n
1
nD0
2

is greater than the general term of our target series. Since all terms are positive, this tells us that
we may conclude the target series converges, by comparison.

˙
 4.2. SERIES CONVERGENCE TESTS 133
The next test is one of the most all-around useful tests. It permits us to resolve any series drawn
from a rational function, for example, by checking to see which p -series a rational function is
most like.

Knowledge Box 4.12

The limit comparison test

Suppose that
1
X 1
X
xn and yn
nD0 nD0

are series, and that

xn
lim DC
n!1 yn

where 0 < jC j < 1 is a constant. Then, either both series converge or both series di-
verge.

Example 4.29 Determine if

X1
nC5
n3C1
nD1

converges or diverges.

Solution:

Perform limit comparison to

X1
1
nD1
n2

nC5
xn n3 C1 n C 5 n2 n3 C 5n2
lim D lim 1
D lim
D lim D1
yn n!1 n!1 n3 C 1 1 n!1 n3 C 1
n2

X1 X1
nC5 1
Since 0 < 1 < 1, we can conclude that 3
converges, because is a convergent
nD1
n C1 nD1
n2
p -series.
˙
134 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Notice that choosing a p -series with p D 2 was exactly what was needed to make the ratio
of the general terms of the series be something that had a finite, non-zero limit. The limit
comparison test is useful for putting rigour into the intuition that one series is “like” another.

The next test uses the fact that if you have a sum of positive terms that is finite, making some or
all of those terms negative cannot take the sum farther from zero than the original finite sum –
leaving the sum finite.

Knowledge Box 4.13

The absolute convergence test

1
X 1
X
If jxn j converges, then so does xn .
nD0 nD0

Example 4.30 Prove that

1 1 1 1
1 C C



4 9 16 25
converges.

Solution:

The series in question is given as an obvious pattern – begin by pulling it into summa-
tion notation:
X1
. 1/nC1
nD1
n2

If we take the absolute value of the general terms of this series, we obtain the series:

X1
1
n2
nD1

We know this to be a convergent p -series. We may deduce that the series converges.

˙
 4.2. SERIES CONVERGENCE TESTS 135
1
X 1
X
Definition 4.9 If jxn j converges, then we say the series xn converges in absolute value.
nD0 nD0

This means we can restate Knowledge Box 4.13 as: “A series that converges in absolute value,
converges.”
The next test uses the fact that if you go up, then down by less, then up by even less, and so on,
you end up a finite distance from your starting point.

Knowledge Box 4.14

The alternating series test

Suppose that xn is a series such that xn > xnC1  0, and suppose that lim xn D 0.
n!1
Then the series
X1
. 1/n
xn
nD0
converges.

X1
. 1/n
Example 4.31 Show that p converges.
nD1
n

Solution:
1
We already know that lim p D 0, and the terms of the series clearly get smaller as n
n!1 n
increases. So, we may conclude this series converges by the alternating series test.

The next two tests both check to see if a series is like a geometric series and deduce its
convergence or divergence from that similarity.
136 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Knowledge Box 4.15

The ratio test

1
X
Suppose that xn is a series. Compute
nD0
ˇ ˇ
ˇ xnC1 ˇ
r D lim ˇˇ ˇ:
n!1 xn ˇ
Then:
• if r < 1, then the series converges,
• if r > 1, then the series diverges,
• if r D 1, then the test is inconclusive.

Definition 4.10 The quantity nŠ D n
.n 1/
.n 2/


2
1 is called n factorial.

We define 0Š D 1.

X1
1
Example 4.32 Determine if converges or diverges.
nD0
nŠ

Solution:

Use the ratio test.

ˇ ˇ ˇ ˇ
ˇ 1=.n C 1/Š ˇ ˇ nŠ ˇ
lim ˇ ˇ D lim ˇ ˇ
n!1 ˇ 1=nŠ ˇ n!1 ˇ .n C 1/Š ˇ

n.n 1/


2
1
D lim
n!1 .n C 1/n.n 1/


2
1

1
D lim D0
n!1 nC1
Since 0 < 1, we can deduce that the series converges by the ratio test.

˙
 4.2. SERIES CONVERGENCE TESTS 137
Knowledge Box 4.16

The root test

1
X
Suppose that xn is a series. Compute
nD0
p
n
s D lim jxn j:
n!1

Then:
• if s < 1, then the series converges,
• if s > 1, then the series diverges,
• if s D 1, then the test is inconclusive.

X1
1
Example 4.33 Determine if converges or diverges.
nD0
nn

Solution:

Use the root test. sˇ ˇ

ˇ1ˇ 1
lim n ˇˇ n ˇˇ D lim D0
n!1 n n!1 n

Since 0 < 1, we can deduce that the series converges by the root test.

The root test and, especially, the ratio test will get a big workout in the next section. This section
contains nine tests for series convergence – many of which depend on knowing the convergence
or divergence behavior of other series. This creates a mental space very similar to the “which
integration method do I use?” issue that arose in Fast Start Integral Calculus. The method for
dealing with this is the same here as it was there: practice, practice, practice.

It’s also good to keep in mind, when you are searching for series to compare to, that the examples
in this section may be used as examples for comparison. A list of series with known behaviors is
138 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
an excellent resource and, given different learning styles, somewhat personal: you may want to
maintain your own annotated list.

4.2.1 TAILS OF SEQUENCES

You may have noticed that we are a little careless with where we start the index of summation
on our infinite series. This is because, while the value of a convergent series depends on every
term, the convergence or divergence behavior does not.

Definition 4.11 If we take a sequence and make a new sequence by discarding a finite number of
initial terms, the new sequence is a tail of the old sequence.

Knowledge Box 4.17

Tail convergence
A sequence converges if and only if all its tails converge.

The practical effect of Knowledge Box 4.17 is that, if a few initial terms of a sequence are
causing trouble, you may discard them and test the remainder of the sequence to determine
convergence or divergence.

Example 4.34 Determine the convergence of the series:

1
X . 1/n
nD0
n2 6n C 10

Solution:

If we look at the first several terms of this series we get:

1 1 1 1 1 1 1
C 1C C C



10 5 2 2 5 10 17

The series alternates signs, getting larger in absolute value for the first four terms, but then
getting smaller in absolute value for the remaining terms. This means that the alternating series
test works for the tail of the sequence starting at the fourth term. Remember that the alternating
series test requires that the terms shrink in absolute value.
 4.2. SERIES CONVERGENCE TESTS 139
We conclude that the series converges by the alternating series test applied to a tail of
the sequence.

The next example shows a very special sort of series for which we can compute the exact value.
Once you understand how these series work, you can construct many examples of them. You
may want to review partial fractions from Fast Start Integral Calculus.

Example 4.35 Prove that the sequence

1
X 1
n2 C 3n C 2
nD0

converges to exactly 1.

Solution:

X1
1
We can see, by limit comparison to , a convergent p -series, that this series converges.
nD1
n2
Knowing the exact value of the sum is another matter. That will require a little algebra.

1
X X1
1 1
2 C 3n C 2
D
nD0
n nD0
.n C 1/.n C 2/

X1  
A B
D C -partial fractions
nD0
nC1 nC2

1 
X 
1 1
D
nD0
nC1 nC2

D1 1=2 C 1=2 1=3 C 1=3 1=4 C 1=4 1=5 C

D 1 C 0 C 0 C 0 C 0 C

D1

˙
140 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
The series in Example 4.35 is called a telescoping series, in analogy to the way a small telescope
or spyglass collapses for storage. In essence, the series is composed of a positive and a negative
series that cancel out all but one term.

Knowledge Box 4.18

Telescoping series
If bn D an anC1 then
1
X
bn D a0 :
nD0

When expanded in terms of the ak values, everything except a0 can-

cels out.

Trace out the information in Knowledge Box 4.18 for Example 4.35.

The next Knowledge Box gives a guide for choosing a convergence test. As with choosing the
best method for integration, the only real way to get better at choosing the correct convergence
test is to work examples. Lots of examples.

Knowledge Box 4.19

Choosing a convergence test

1. If the general term of the series fails to converge to zero, then it
diverges by the divergence test. Remember that this does not work
in reverse – if the general term converges to zero, anything might
happen.
2. If the series is a geometric series or a p -series, use the tests for those
series.
3. If the general term of the series is an D f .n/ for a function f .x/
that you can integrate, try the integral test.
4. A positive series that is, term-by-term, no larger than a series that
converges, converges (comparison test) – look for this.

continued
 4.2. SERIES CONVERGENCE TESTS 141
Knowledge Box 4.20

Choosing a convergence test—Continued

5. Similarly, a positive series that is, term-by-term, no smaller than
a divergent series diverges, again by the comparison test.
6. If you have a series that looks like a series you know how to deal
with, try the limit comparison test. This is useful for things like
slightly modified p series or letting you use simpler integral tests.
7. You can often set up a comparison test or limit comparison test by
doing algebra or arithmetic to the general term of a series.
8. Remember that if the term-by-term absolute value of a series con-
verges, then the series converges. This is the absolute convergence
test.
9. If the terms of a series alternate in sign, look at the alternating series
test.
10. If you can take the limit of adjacent terms of a series in a reasonable
way, then the ratio test is a possibility.
11. If you can take the nth root of the general term of a series in a rea-
sonable way, the root test is a possibility.
12. If some finite number of initial terms of a series are preventing you
from using a test, tail convergence says you can ignore them and
then do your test.
13. Unless you’re taking a test or quiz, asking for advise and suggestions
is not the worst possible option.
142 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
PROBLEMS

Problem 4.36 For each of the following series, determine if the series converges or diverges.
State the name of the test you are using.

X1 X1 X1
1 nC1 2n C 1
1. e
3. 5.
nD1
n nD0
5n C4 nD0
3n C 1

X1 1
X   X1 p
nC1 1 n
2. 2C1
4. ln 6.
nD0
n nD1
n2 nD0
n2C1

Problem 4.37 For each of the following series, determine if the series converges or diverges.
State the name of the test you are using.

1
X 1
X X1
n2 sin.n/ . 1/n
1. n
3. p 5. p3
nD0
2 C1 nD2 n 2
nD1
n
1
X X1 1
X
1 1
2. 0:0462n 4. p
3
6.
nD0 nD1
n nD0
.2n/Š

Problem 4.38 For each of the following series, determine if the series converges or diverges.
State the name of the test you are using.

X1 1
X X1
1 n e2n
1. 3. e 5.
nD1
nn=2
nD1 nD1
n

X1 X1 1
X
5n en n2
2. 4. 6. e3Cn
nD1
nn=2 nD1
n nD1

Problem 4.39 Give an example to demonstrate that the divergence test does not work in re-
verse, i.e., a sequence whose general term goes to zero but whose sum is infinite.

Problem 4.40 Use the integral test to prove that the p -test works.
 4.2. SERIES CONVERGENCE TESTS 143
Problem 4.41 Suppose that p.x/ is a polynomial. Use the integral test to demonstrate that
1
X
n
p.n/e
nD0

converges.

Problem 4.42 Suppose that q.x/ is a polynomial with exactly three roots, all of which are
negative real numbers. Demonstrate that

X1
1
nD0
q.n/

converges.

Problem 4.43 If xn and yn are the general terms of a convergent series, then xn C yn are as
well. This requires only simple algebra. What is startling is that the reverse is not true. Find an
example of an D xn C yn so that
X1
an
nD1

converges, but neither of

1
X 1
X
xn or yn
nD1 nD1
converge.

Problem 4.44 Show that, when you apply the ratio test to a geometric series, the limit that
appears in the test is the ratio of the series.

Problem 4.45 Show that, when you apply the root test to a geometric series, the limit that
appears in the test is the ratio of the series.

Problem 4.46 Suppose r > 1. Prove that

X1
nk
nD0
rn

converges when k is an integer  1.

144 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Problem 4.47 Suppose that we are testing the convergence of a series
X1
p.n/
nD0
q.n/

where p.x/ and q.x/ are polynomials. If we add the assumption that q.x/ has no roots at any
of the values of n involved in the sum, explain in terms of the degrees of the polynomials when
the series converges.

Problem 4.48 For each of the following series, determine if the series converges or diverges.
State the name of the test you are using.

1
X X1 p
n3 n
1. 4. 2C1
nD0
.1 C n/.2 C n/.3 C n/.4 C n/ nD0
n
1
X 1
X
n2 n
2. 5. 1:25
nD0
.1 C n/.2 C n/.3 C n/.4 C n/ nD0

X1 1
X
n5
3. 6. e n

nD0
en nD0

1
X 1
Problem 4.49 Compute exactly: 2Cn
nD1
n

1
X 1
Problem 4.50 Compute exactly:
nD1
n2 C 2n

1
X 1
Problem 4.51 Compute exactly: 2 C 5n
nD1
n

X1
sin.n/
Problem 4.52 Demonstrate convergence of the series
nD1
n2 C n

1
X 1
Problem 4.53 Compute exactly:
nD0
4n2 C 8n C 3
 4.3. POWER SERIES 145
4.3 POWER SERIES
In this section we study series again. The good news is that we do not have any additional
convergence tests. The bad news is that these series will have variables in them.

Definition 4.12 A power series is a series of the form:

1
X
an x n
nD0

In a way, a power series is actually an infinite number of different ordinary series, one for each
value of x you could substitute into it. The goal of this section will be: given a power series, find
values of x which cause it to converge.

Knowledge Box 4.21

The radius of convergence of a power series

The power series
1
X
an x n
nD0

converges in one of three ways:

1. Only at x D 0.
2. For all jxj < r and possibly at x D ˙ r .
3. For all x .

The number r is the radius of convergence of the power series. In the first case
above, we say the radius of convergence is zero; in the third, we say the radius of
convergence is infinite.

Definition 4.13 The interval of convergence of a power series is the set of all x where it converges.

Knowledge Box 4.21 implies that the interval of convergence of a power series is one of Œ 0; 0 ,
. r; r/, Œ r; r/, . r; r, Œ r; r, or . 1; 1/. The results with an r in them occur in the case
146 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
where 0 < r < 1. Once we have the radius of convergence, in the case where r is positive and
finite, we determine the interval of convergence by checking the behavior of the series when we
set x D ˙ r .

Example 4.54 Find the radius and interval of convergence of:

1
X
xn
nD0

Solution:

We start by trying to determine the radius of convergence.

Use the ratio test:

ˇ nC1 ˇ
ˇx ˇ
lim ˇˇ n ˇˇ D lim jxj D jxj
n!1 x n!1

This is true because x does not depend on n.

The series thus converges when jxj < 1, meaning we have convergence for sure when
1 < x < 1.

This also means the radius of convergence is r D 1.

We now need to check x D ˙1 to determine the interval of convergence.

These values both yield non-converging geometric series:

1
X 1
X
n
. 1/ and 1
nD0 nD0

So the potential endpoints of the interval of convergence are not part of the interval of
convergence. This means that the interval of convergence is . 1; 1/.

˙
 4.3. POWER SERIES 147
Example 4.55 Find the radius of convergence of:

X1
xn
nD0
2n

Solution:

Again, use the ratio test.

ˇ ˇ ˇ nC1 nC1 ˇ
ˇ anC1 ˇ ˇx =2 ˇ
ˇ
lim ˇ ˇ D lim ˇ ˇ
n!1 a ˇ n!1 ˇ x n =2n ˇ
n

ˇ nC1 ˇ
ˇx 2n ˇˇ
ˇ
D lim ˇ nC1
n ˇ
n!1 2 x
ˇx ˇ jxj
ˇ ˇ
D lim ˇ ˇ D
n!1 2 2

So the series converges when:

x
1< <1
2

2<x<2

The radius of convergence is r D 2.

Example 4.56 Find the radius of convergence of the series:

X1
xn
nD0
nŠ
148 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Solution:

Like before:
ˇ ˇ ˇ nC1 ˇ
ˇ anC1 ˇ ˇx =.n C 1/Š ˇ
lim ˇˇ ˇ D lim ˇ
ˇ ˇ n
ˇ
ˇ
n!1 an n!1 x =nŠ
ˇ ˇ
ˇ x ˇ
D lim ˇˇ ˇ
n!1 n C 1 ˇ

ˇ ˇ
ˇ 1 ˇ
ˇ
D jxj lim ˇ ˇ
n!1 n C 1 ˇ

D jxj
0

D0

Since 0 < 1 for all values of x , this power series converges everywhere and the radius of
convergence is infinite.

Example 4.57 Find the radius of convergence of the series:

X1
xn
nD0
nn

Solution:

This time use the root test.

sˇ ˇ
ˇ xn ˇ ˇ ˇ
lim n ˇ ˇ D lim ˇˇ x ˇˇ
n!1 ˇ nn ˇ n!1 n

1
D jxj
lim D0
n!1 n

Which tells us that, as 0 < 1 for all x , that this sequence converges everywhere.

˙
 4.3. POWER SERIES 149
Example 4.58 Find the interval of convergence of the series:

X1
xn
nD1
n2

Solution:

Another natural job for the ratio test.

ˇ ˇ ˇ nC1 ˇ
ˇ anC1 ˇ ˇx =.n C 1/2 ˇˇ
lim ˇ ˇ ˇ
D lim ˇ
n!1 ˇ an ˇ n!1 x n =n2 ˇ

ˇ ˇ
ˇ n2 ˇ
D lim ˇˇ
x ˇ
ˇ
n!1 .n C 1/2

n2
D jxj
lim
n!1 n2 C 2n C 1

D jxj
1 D jxj

So 1 < x < 1, and the radius of convergence is r D 1. To find the interval of convergence we
need to check the ends of the interval. These are of the form

X1
.˙1/n
nD1
n2

The absolute value of each of these is a p -series with p D 2. Both endpoints correspond to
series that converge in absolute value – meaning they both converge. This makes the interval of
convergence Œ 1; 1.

4.3.1 USING CALCULUS TO FIND SERIES

It is sometimes useful to represent functions as power series. The geometric series formula
1
(Knowledge Box 4.7) does this, for example, for the function f .x/ D . We can use calculus
1 x
to derive power series for other functions.
150 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Example 4.59 If jxj < 1, then the geometric series formula tells us that:
1
X 1
xn D
nD0
1 x

If we plug x into this identity we find that:

1
X 1
. x/n D
nD0
1 . x/
1
X 1
. 1/n x n D
nD0
1Cx
or
1
D 1 x C x2 x3 C x4



xC1
Integrate both sides and we get:
1 2 1 3 1 4 1 5
ln.x C 1/ C C D x x C x x C x



2 3 4 5
Plug in x D 0
ln.1/ C C D 0
C D0
Which means, at least when jxj < 1,
X1
. 1/nC1 x n
ln.x C 1/ D
nD1
n
˙
This shows that we can find power series that, when they converge, are equal to familiar tran-
scendental functions. The next section gives another technique for doing this, to be used when
the sequences you already know don’t give you enough power. Let’s encode this as a Knowledge
Box.
Knowledge Box 4.22

Using calculus to modify power series

1
X
Suppose that f .x/ D an x n . Then:
nD0

Z X1 X1
an nC1 0
f .x/
dx D x and f .x/ D n
an x n 1 :
nD0
nC1 nD1
 4.3. POWER SERIES 151
1
Example 4.60 Find a power series for tan .x/.
Solution:

We already have seen that:

1
D1 u C u2 u3 C u4



uC1
If we substitute u D x 2 we obtain:
1
D1 x2 C x4 x6 C x8



x2 C1
Integrate and we get:

1 1 3 1 5 1 7 1 9
tan .x/ C C D x x C x x C x



3 5 7 9
Substitute in x D 0 and we get C D 0. So we obtain the power series:
X1
1 . 1/n x 2nC1
tan .x/ D
nD0
2n C 1

Example 4.61 What is the interval of convergence for the series for
1
f .x/ D tan .x/

found in Example 4.60?

Solution:

Apply the ratio test:

ˇ ˇ ˇ 2nC3 ˇ
ˇ anC1 ˇ ˇ =.2n C 3/ ˇˇ
lim ˇˇ ˇ D lim ˇ x
n!1 an ˇ n!1 ˇ x 2nC1 =.2n C 1/ ˇ
ˇ ˇ
ˇ 2 2n C 1 ˇ
D lim ˇˇx ˇ
n!1 2n C 3 ˇ
2n C 1
D jx 2 j lim
n!1 2n C 3

D jx 2 j
1 D x 2
152 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
So x 2 < 1 when 1 < x < 1, making the radius of convergence r D 1.

Now check the endpoints x D ˙1.

If x D 1 the resulting series is:

Xn Xn
. 1/n . 1/2nC1 . 1/nC1
D
nD0
2n C 1 nD0
2n C 1

which converges by the alternating series test.

If x D 1 we get:
Xn
. 1/n
nD0
2n C 1
which also converges by the alternating series test.

The interval of convergence is thus Œ 1; 1.

˙

It is possible to find a power series by just using algebra on a known series.

Example 4.62 Find a power series for:

x2
f .x/ D
1 x2

Solution:

We start with the known form:

1
D 1 C u C u2 C u3 C u4 C



1 u
Substitute in u D x 2 and we get:
1
D 1 C x2 C x4 C x6 C x8 C



1 x2
Now multiply both sides by x 2 and we get:
x2
D x 2 C x 4 C x 6 C x 8 C x 10 C



1 x2
 4.3. POWER SERIES 153
So
1
X
f .x/ D x 2nC2
nD0

Example 4.63 Find a power series for:

g.x/ D ln.x 2 C 1/

Solution:
1
Start with the known result for .
1 C x2

1
D1 x2 C x4 x6 C x8



1 C x2

2x
D 2x 2x 3 C 2x 5 2x 7 C 2x 9



1 C x2
Z Z
2x


dx D 2x 2x 3 C 2x 5 2x 7 C 2x 9


dx
1 C x2
 
2 1 2 1 4 1 6 1 8 1
ln.x C 1/ C C D 2 x x C x x C x 10



2 4 6 8 10

Set x D 0 and we get C D 0:

X1
2
. 1/n x 2nC2
So: ln.x 2 C 1/ D
nD0
2n C 2

˙
154 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
An integral that is both important in statistics, because it is related to the normal distribution,
and famous for not having a closed form solution is:
Z
2
e x =2
dx

The next example shows why power series are useful – it is easy to get a power series for this
integral.

Example 4.64 Find a power series for

Z
x 2 =2
F .x/ D e
dx

Solution:

X1
xn
ex D This will be shown in Section 4.4
nD0
nŠ

1
n
X x 2 =2
x 2 =2
e D
nD0
nŠ

X1
x 2 =2 . 1/n x 2n
e D
nD0
2n
nŠ

Z Z X1
x 2 =2 . 1/n x 2n
e
dx D
dx
nD0
2n
nŠ

Z 1
X
x 2 =2 . 1/n x 2nC1
e
dx D
nD0
.2n C 1/
2n
nŠ

This power series is useful for computing probabilities connected with the normal distribution.

˙
 4.3. POWER SERIES 155
PROBLEMS

Problem 4.65 Find the radius of convergence for each of the following power series.

X1 X1 1
X
xn x 2nC1 xn
1. 4. 7.
nD1
n nD0
nŠ nD0
2n2 C 4

X1 1
X 1
xn X xn
2. 5. nx n 8.
nD0
5n nD1 n3 C 1
nD0
1
X X1
xn nx n
3. 6.
nD0
.2n C 1/Š nD0
3n C 1

Problem 4.66 Demonstrate that the radius of convergence of

1
X 1
X
an x n and can x n
nD0 nD0

are the same for any constant c .

Problem 4.67 Compute the radius of convergence of

X1
xn
nD1
cn

where c > 0 is a constant.

Problem 4.68 If
1 1 1
f .x/ D C C
1 x 1 2x 1 3x
find and simplify a power series for f .x/.

Problem 4.69 If
x x
g.x/ D C
2 x 3 x
find and simplify a power series for g.x/.
156 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Problem 4.70 If
x2 x2
h.x/ D C
2 x 1 2x
find and simplify a power series for h.x/.

Problem 4.71 Find the interval of convergence for each of the following power series.

X1 1
X X1
xn nx n xn
1. 3. 5.
nD0
4n nD0
2n3 C 7 nD0
nn
1
X X1 X1
xn n2 x n . 2x/n
2. 4. 6.
nD1
.2n C 1/Š nD2
n4 1 nD0
n

Problem 4.72 Using calculus, find a power series for:

 
1Cx
g.x/ D ln
1 x

Problem 4.73 Using only algebra, and known series, find a power series for:

3x 2
h.x/ D
1 C x3

Problem 4.74 Find a power series for:

q.x/ D ln 1 C x 3

Problem 4.75 Find a power series for:

1
q.x/ D
x2 9
 4.4. TAYLOR SERIES 157
Problem 4.76 Find a power series for:
1
q.x/ D
x2 3x C 2

Problem 4.77 Find a power series for:

q.x/ D x 2
ln 1 C x 3

Problem 4.78 Find a power series for:

x2 C 9
q.x/ D
x2 9

4.4 TAYLOR SERIES

In the last section, we managed to create power series for several of the standard transcendental
functions. Notably absent were sin.x/, cos.x/, and ex . The key to these is Taylor series. We
need a little added notation to build Taylor series. We will denote the nth derivative of f .x/ by
f .n/ .x/. Notice that this means that f .0/ .x/ D f .x/.

Knowledge Box 4.23

Taylor series
If f .x/ is a function that can be differentiated any number of times,
1
X f .n/ .c/.x c/n
f .x/ D :
nD0
nŠ

This formula is called the Taylor series expansion of f .x/ at c. The constant c is
called the center of the expansion.

Example 4.79 Use Taylor’s formula to find a power series centered at c D 0 for f .x/ D ex and
find its radius of convergence.
158 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Solution:

The function f .x/ D ex is a very good choice for a first demonstration of the Taylor
expansion. This is because every derivative of of ex is ex . In other words,

f .n/ .x/ D ex and so f .n/ .0/ D 1

Applying the formula we get:

X1 X1 X1
f .n/ .0/.x 0/n 1
xn xn
ex D D D
nD0
nŠ nD0
nŠ nD0
nŠ

The radius of convergence of this series was computed in Example 4.56 – it is r D 1. This
means that the expansion of ex converges everywhere.

˙
Here is an interesting calculation. Recall Euler’s identity from Fast Start Differential Calculus:
eix D i sin.x/ C cos.x/. Let’s look at the Taylor series for eix :

X1
.ix/n
eix D
nD0
nŠ

X1 n n
i x
D
nD0
nŠ

1 2 1 3 1 1 1 6 1 7
D 1 C ix x i x C x4 C i x5 x i x C



2 3Š 4Š 5Š 6Š 7Š

X1 X1
. 1/n x 2n . 1/n x 2nC1
D Ci
nD0
.2n/Š nD0
.2n C 1/Š

From Euler’s identity, we get that the real part of the expression above is cosine, and the imag-
inary part is sine. This gives us power series for sin.x/ and cos.x/.
 4.4. TAYLOR SERIES 159
Knowledge Box 4.24

Taylor series for sin.x/ and cos.x/ and ex

1
X
x xn
e D
nD0
nŠ
X1
. 1/n x 2nC1
sin.x/ D
nD0
.2n C 1/Š
X1
. 1/n x 2n
cos.x/ D
nD0
.2n/Š
with all three expansions having an interval of convergence of . 1; 1/.

The calculations above show that we can use algebraic manipulation to create power series based
on the power series we get from Taylor expansions. Let’s do a couple more examples.

Example 4.80 Find a power series for h.x/ D e2x .

Solution:

X1
xn
ex D
nD0
nŠ

X1
.2x/n
e2x D
nD0
nŠ

X1
2n x n
D
nD0
nŠ

X1
2n n
D x
nD0
nŠ

˙
160 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION

Example 4.81 Find the Taylor expansion for s.x/ D cos.x/ using c D .
2

Solution:

The Taylor formula needs the nth derivatives at . Let’s start by computing these num-
2
bers.


n s .n/ .x/ s .n/ 2
0 cos.x/ 0
1 sin.x/ -1
2 cos.x/ 0
3 sin.x/ 1
4 cos.x/ 0
5 sin.x/ -1

Which is enough to notice the values repeat every four steps.

Plug these values into the Taylor expansion formula with c D =2 and we get:

.x =2/ .x =2/3 .x =2/5 .x =2/7

cos.x/ D C C
1Š 3Š 5Š 7Š

X1
. 1/nC1 .x =2/2nC1
cos.x/ D
nD0
.2n C 1/Š

˙

Notice that the expansion of cos.x/ with a center of c D has the same form as the negative
2
of the Taylor expansion of sin.x/ at c D 0. This is a fairly extreme way of proving the identity

cos.x/ D sin.x =2/

4.4.1 TAYLOR POLYNOMIALS

If we take the first n terms of a Taylor series for f .x/ the result is called the Taylor polynomial of
degree n for f .x/. Taylor polynomials are approximations to the function they are derived from
and, as we already know, polynomials are among the very nicest functions to work with.
 4.4. TAYLOR SERIES 161
Example 4.82 Find a fifth-degree Taylor polynomial for f .x/ D sin.x/ centered at c D 0.
Solution:

Since the Taylor series for sin.x/ centered at c D 0 is

X1
. 1/n x 2nC1
nD0
.2n C 1/Š

we need only extract the terms of degree five or less from the infinite series. This gives us a
solution of
x3 x5
p.x/ D x C
6 120
˙

A natural question is “how good is this polynomial as an approximation to sin.x/?

Let’s graph both functions on the same set of axes.

2
x3 x5
p.x/ D x 6
C 120

-4 4
(0,0)
f .x/ D sin.x/

-2

In the range 2  x  2 the polynomial and the sine function agree really well – after that they
diverge from one another and the polynomial shoots off to positive and negative infinity. Not
too surprisingly, the Taylor polynomial does a good job of approximating the function it was
drawn from near the center c for expansion. How do we figure out where the polynomial is a
good enough approximation?
If T .x/ is the Taylor series for a function f .x/ centered at c and Tn .x/ is the nth degree Taylor
polynomial for that function, then we set Rn .x/ D T .x/ Tn .x/ to get the remainder for the
162 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
polynomial. Algebraically rearranging the terms:

T .x/ D Tn .x/ C Rn .x/

With this definition we can give Taylor’s inequality for the remainder of a power series.

Knowledge Box 4.25

Taylor’s Inequality
Suppose we are looking at a Taylor polynomial for the function f .x/ in the inter-
val jx aj  d and that jf .nC1/ .x/j  M everywhere in this interval. Then
for values of x in the interval we have that
M
jRn .x/j  jx ajnC1 :
.n C 1/Š

It is not immediately obvious how to use this result, so let’s do a couple examples.

Example 4.83 Find a bound on the error of approximation of T5 .x/ for f .x/ D sin.x/ on the
interval 2  x  2. Note that this is the interval that looked good in the graph associated with
Example 4.82.
Solution:

We proceed by applying Taylor’s inequality. The derivative of sin.x/ are all sine and cosine
functions. This means that f .nC1/ .x/ are always at most 1. So, we may set M D 1 for Taylor’s
inequality. The value of a is zero so jx aj  2 on the interval we are using.

Plugging these values into the formula for Taylor’s inequality we get:

1 6 64
jR5 .x/j 
2 D Š 0:089
6Š 720
The error of T5 .x/ on 2  x  2 is at most 0.089. Not bad.

Notice that Taylor’s inequality has .n C 1/Š in the denominator. That means, if we use

x3 x5 x7
T7 .x/ D x C
6 120 5040
 4.4. TAYLOR SERIES 163
for f .x/ D sin.x/ on the interval 2  x  2, then the estimate of maximum error drops to

1 8 256
jR7 .x/j 
2 D Š 0:0063
8Š 40320
The factorial in the denominator lets the error drop really quickly.

The hard part of using Taylor’s inequality is finding the constant M . Standard practice is to
simply find the largest value of f nC1 .x/ in the interval and live with it. The fact that both sine
and cosine are bounded in absolute value by 1 make them especially easy functions to work with.
Let’s do an example with an exponential function.

Example 4.84 Suppose we are approximating f .x/ D ex in the interval 3  x  3 with

Tn .x/. What value of n makes the Taylor’s inequality estimate of error at most 0.1?
Solution:

If f .x/ D ex then f .nC1/ D ex , which is very convenient. We again have that a D 0. So, it is
pretty easy to see that

jf .nC1/ .x/j  e3

everywhere on the interval. So, we set M D e3 . The largest value of jx aj on the interval is 3,
so we need to find the smallest value of n that makes

e3
3nC1 < 0:1
.n C 1/Š

The simplest way to do this is to tabulate.

e 3 e3
n jRn .x/j  j .nC1/Š
3nC1 j n jRn .x/j  j .nC1/Š
3nC1 j
1 90.3849161543 6 8.7156883435
2 90.3849161543 7 3.2683831288
3 67.7886871158 8 1.0894610429
4 40.6732122695 9 0.3268383129
5 20.3366061347 10 0.0891377217
164 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
So the first value of n with an acceptable error is n D 10, and T10 .x/ is good enough to
approximate f .x/ D ex in the range 3  x  3.

The error estimates given by Taylor’s inequality are not the best possible – they are actually fairly
conservative. They usually over-estimate the error. If you take a course in numerical analysis
later in your career you may study methods for building better error estimates. There is also a
lot of room to be clever with how you use Taylor polynomials. The sine and cosine function are
periodic, and so if you know their values even on a very small interval, like Œ0; =2, you can use
those values to deduce any other value.

PROBLEMS

Problem 4.85 Using the Taylor series formula, verify the formula for y D sin.x/.

Problem 4.86 Using the Taylor series formula, verify the formula for y D cos.x/.

Problem 4.87 Find the Taylor expansion of f .x/ D sin.x/ using a center of c D  . Use the
formula for Taylor expansion.

Problem 4.88 Find the Taylor expansion of f .x/ D sin.x/ using a center of c D =2. Use the
formula for the Taylor expansion.

Problem 4.89 Using any method, find power series for the following functions.

x
1. f .x/ D e 5. s.x/ D ln.x 4 /

2. g.x/ D xe2x 6. q.x/ D .ex C e x

/=2

7. a.x/ D tan 1 .3x/

3. h.x/ D cos 2x
3x 2
4. r.x/ D sin.x 2 / 8. b.x/ D
1 x4
 4.4. TAYLOR SERIES 165
Problem 4.90 Find the Taylor expansion of f .x/ D cos.x/ using a center of c D =4. Use the
formula for the Taylor expansion. Warning: this is a little messy.

Problem 4.91 Using the Taylor series for sin.x/, cos.x/ and ex , prove Euler’s identity:

ei D i sin.x/ C cos.x/

Problem 4.92 For each of the series you found in Problem 4.89, find the radius and interval
of convergence.

Problem 4.93 Prove that the Taylor series for a polynomial function p.x/ is just the polynomial
itself.

Problem 4.94 Find a power series expansion for

1
f .x/ D
x2 3x C 2

Problem 4.95 Find a power series expansion for

1
f .x/ D
4 4x C x 2

Problem 4.96 Find a power series expansion for

1
f .x/ D
x3 6x 2 C 11x C 6

Problem 4.97 Find a power series expansion for

1
f .x/ D
x3 Cx
166 4. SEQUENCES, SERIES, AND FUNCTION APPROXIMATION
Problem 4.98 If
f .x/ D p.x/ex
where p.x/ is a polynomial, demonstrate that f .x/ has a power series expansion with radius of
convergence r D 1.

Problem 4.99 Find the Taylor polynomial of degree n for the given function with the given
center c .

1. f .x/ D cos.x/ for n D 6 at c D 0, 4. r.x/ D log.x/ for n D 3 at c D 1,

2. g.x/ D sin.2x/ for n D 7 at c D 0, 5. s.x/ D tan 1 .x/ for n D 8 at c D 0,
3. h.x/ D ex for n D 5 at c D 0, 6. q.x/ D x 2 C 3x C 5 for n D 2 at c D 1.

Problem 4.100 Suppose we have T5 .x/ for f .x/ D ex at c D 0. Compute a bound on the size
of R5 .x/ with Taylor’s inequality.

Problem 4.101 Find the smallest n for which Tn .x/ on f .x/ D cos.x/ has jRn .x/j < 0:01 on
3  x  3 with c D 0.
 167

APPENDIX A

Useful Formulas
A.1 POWERS, LOGS, AND EXPONENTIALS
RULES FOR POWERS

n 1 an
• a D • D an m
an am
• an  am D anCm • .an /m D anm

LOG AND EXPONENTIAL ALGEBRA

• b logb .c/ D c • logb .x y / D y
logb .x/

• logb .b a / D a
logb .x/
• logb .xy/ D logb .x/ C logb .y/ • logc .x/ D
  logb .c/
x
• logb D logb .x/ logb .y/
y • If logb .c/ D a, then c D b a

A.2 TRIGONOMETRIC IDENTITIES

TRIG FUNCTION DEFINITIONS FROM SINE AND COSINE

sin./ 1 1
• tan./ D • tan. / D • csc. / D
cos./ cot. / sin. /

cos./ 1
• cot. / D • sec./ D
sin. / cos./

PERIODICITY IDENTITIES



• sin.x C 2/ D sin.x/ • sec.x/ D csc x C 2
• sin.x C / D sin.x/
• cos.x C 2/ D cos.x/ • cos. x/ D cos.x/ • cos.x C / D cos.x/


• sin.x/ D cos x 2 • sin. x/ D sin.x/


• tan.x C / D tan.x/
• tan.x/ D cot x 2 • tan.x/ D tan.x/
168 A. USEFUL FORMULAS
THE PYTHAGOREAN IDENTITIES

• sin2 ./ C cos2 ./ D 1 • tan2 . / C 1 D sec2 ./ • 1 C cot2 ./ D csc2 . /

THE LAW OF SINES, THE LAW OF COSINES

B A

˛ ˇ
C
The Law of Sines The Law of Cosines

A B C
D D
sin.˛/ sin.ˇ/ sin. / C 2 D A2 C B 2 C 2AB
cos. /

The laws refer to the diagram.

SUM, DIFFERENCE, AND DOUBLE ANGLE

• sin.˛ C ˇ/ D sin.˛/ cos.ˇ/ C sin.ˇ/ cos.˛/ • cos.2/ D cos2 . / sin2 ./

• cos.˛ C ˇ/ D cos.˛/ cos.ˇ/ sin.˛/ sin.ˇ/
1 C cos. /
• sin.˛ ˇ/ D sin.˛/ cos.ˇ/ sin.ˇ/ cos.˛/ • cos2 .=2/ D
2
• cos.˛ ˇ/ D sin.˛/ sin.ˇ/ C cos.˛/ cos.ˇ/
1 cos. /
• sin.2/ D 2 sin./ cos./ • sin2 .=2/ D
2

A.3 SPEED OF FUNCTION GROWTH

• Logarithms grow faster than constants.
• Positive powers of x grow faster than logarithms.
• Larger positive powers of x grow faster than smaller positive powers of x .
• Exponentials (with positive exponents) grow faster than positive powers of x .
• Exponentials with larger exponents grow faster than those with smaller exponents.
 A.4. DERIVATIVE RULES 169
A.4 DERIVATIVE RULES

• If f .x/ D x n then • .sec.x//0 D sec.x/ tan.x/

f 0 .x/ D nx n 1
• .csc.x//0 D csc.x/ cot.x/

0 1
• .f .x/ C g.x//0 D f 0 .x/ C g 0 .x/ • sin 1 .x/ D p
1 x2
• .a
f .x//0 D a
f 0 .x/

0 1
1 • cos 1 .x/ D p
• If f .x/ D ln.x/, then f 0 .x/ D 1 x2
x
1
0 1
• If f .x/ D logb .x/, then f 0 .x/ D • tan 1 .x/ D
x ln.b/ 1 C x2
• If f .x/ D ex , then f 0 .x/ D ex

0 1
• cot 1 .x/ D
• If f .x/ D ax , then f 0 .x/ D ln.a/
ax 1 C x2
• .sin.x//0 D cos.x/
0 1
0
• sec 1 .x/ D p
• .cos.x// D sin.x/ jxj x 2 1

• .tan.x//0 D sec2 .x/
0 1

• csc 1 .x/ D p
• .cot.x//0 D csc2 .x/ jxj x 2 1

The product rule

.f .x/
g.x//0 D f .x/g 0 .x/ C f 0 .x/g.x/

The quotient rule

 0
f .x/ g.x/f 0 .x/ f .x/g 0 .x/
D
g.x/ g 2 .x/

The reciprocal rule

 0
1 f 0 .x/
D
f .x/ f 2 .x/

The chain rule

.f .g.x///0 D f 0 .g.x//
g 0 .x/
170 A. USEFUL FORMULAS
A.5 SUMS AND FACTORIZATION RULES
FACTORIZATIONS OF POLYNOMIALS
• x 2 a2 D .x a/.x C a/

• x3 a3 D .x a/.x 2 C ax C a2 /

• x 3 C a3 D .x C a/.x 2 ax C a2 /

• xn an D .x a/.x n 1
C ax n 2
C


an 2
x C an 1
/

ALGEBRA OF SUMMATION

b
X b
X b
X b
X b
X
• f .i / C g.i / D f .i / C g.i / • c
f .i/ D c
f .i/
i Da i Da i Da i Da i Da

CLOSED SUMMATION FORMULAS

n
X n
X n.n C 1/.2n C 1/
• 1Dn • i2 D
6
i D1 i D1

n
X n
X
n.n C 1/ n2 .n C 1/2
• iD • i3 D
2 4
i D1 i D1

A.5.1 GEOMETRIC SERIES

n
X 1
X
anC1 1 q
• k
a D • qan D if jaj < 1
a 1 nD0
1 a
kD0
1
X 1
• an D if jaj < 1
nD0
1 a
 A.6. VECTOR ARITHMETIC 171
A.6 VECTOR ARITHMETIC
VECTOR ARITHMETIC AND ALGEBRA
• c
vE D .cv1 ; cv2 ; : : : ; cvn / • c
.E
v C w/
E D c
vE C c
w
E
• vE C wE D .v1 C w1 ; v2 C w2 ; : : : ; vn C wn / • c
.d
vE/ D .cd /
vE
• vE w
E D .v1 w1 ; v2 w2 ; : : : ; vn wn / • vE C wE D wE C vE
• vE
wE D v1 w1 C v2 w2 C : : : C vn wn • uE
.E
v C w/
E D uE
vE C uE
w
E

CROSS PRODUCT OF VECTORS

• vE  wE D . v2 w3 v3 v2 ; v3 w1 v1 w3 ; v1 w2 v2 w1 /

Formula for the angle between vectors

vE
w
E
cos./ D
jvjjwj

A.7 POLAR AND RECTANGULAR CONVERSION

p
• x D r
cos./ • rD x2 C y2

• y D r
sin. / •  D tan 1 .y=x/

A.8 INTEGRAL RULES

BASIC INTEGRATION RULES
Z Z Z
n 1 nC1
• x
dx D x CC • a
f .x/
dx D a
f .x/
dx
nC1
Z Z Z
• .f .x/ C g.x//
dx D f .x/
dx C g.x/
dx

LOG AND EXPONENT

Z Z
1
•
dx D ln.x/ C C • ex
dx D ex C C
x
172 A. USEFUL FORMULAS
TRIG AND INVERSE TRIG
Z
• sin.x/
dx D cos.x/ C C Z
1
Z • p
dx D sin 1 .x/ C C
1 x 2
• cos.x/
dx D sin.x/ C C Z
1
Z •
dx D tan 1 .x/ C C
1 C x2
• sec2 .x/
dx D tan.x/ C C Z
1
Z • p
dx D sec 1 .jxj/ C C
• csc2 .x/
dx D cot.x/ C C x x2 1
Z
Z • tan.x/
dx D ln j sec.x/j C C
• sec.x/ tan.x/
dx D sec.x/ C C Z
Z • sec.x/
dx D ln j sec.x/ C tan.x/j C C
• csc.x/ cot.x/
dx D csc.x/ C C

INTEGRATION BY PARTSZ Z
U
dV D U V V
dU

EXPONENTIAL/POLYNOMIAL
Z SHORTCUT


p.x/ex
dx D p.x/ p 0 .x/ C p 00 .x/ p 000 .x/ C


ex C C

VOLUME, SURFACE, ARC LENGTH

Volume of Rotation, Disks Differential of Arc Length

Z b q q
V D f .x/2
dx ds D .y /
dx D .f 0 .x//2
dx
0 2
a

Volume of Rotation, Cylindrical Shells Arc Length

Z Z
xDb
V D 2 x
f .x/
dx SD ds
xDa

Volume of Rotation with Washers Surface Area of Rotation

Z b Z b
2 2


V D f1 .x/ f2 .x/
dx A D 2 f .x/
ds
a a
 A.9. SERIES CONVERGENCE TESTS 173
A.9 SERIES CONVERGENCE TESTS
DIVERGENCE TEST
1
X
If lim xn ¤ 0 then xn does not have a finite value.
n!1
nD0

INTEGRAL TEST
1
X
Suppose that f .x/ is a positive, decreasing function on Œ0; 1/ and that xn D f .n/. Then xn
Z 1 nD1

and f .x/dx both converge or both diverge for any finite a  1.

a

P-SERIES TEST
X1
1
The p -series p
converges if p > 1 and diverges if p  1.
nD1
n

COMPARISON TESTS
1
X 1
X
• If converges and 0  yn  xn for all n, then yn also converges.
nD0 nD0

1
X 1
X
• If xn diverges and yn  xn  0 for all n, then yn also diverges.
nD0 nD0

LIMIT COMPARISON TEST

1
X 1
X xn
Suppose that xn and yn are series, and lim
D C where 0 < jC j < 1 is a constant.
nD0 nD0
ynn!1

Then, either both series converge or both series diverge.

ABSOLUTE CONVERGENCE TEST

1
X 1
X
If jxn j converges, then so does xn .
nD0 nD0

ALTERNATING SERIES TEST

Suppose that xn is a series such that xn > xnC1  0, and suppose that lim xn D 0. Then the
n!1
1
X
n
series . 1/
xn converges.
nD0
174 A. USEFUL FORMULAS
RATIO TEST
1
X ˇ ˇ
ˇ xnC1 ˇ
Suppose that ˇ
xn is a series and r D lim ˇ ˇ. Then
n!1 x ˇ
nD0 n

• if r < 1, then the series converges

• if r > 1, then the series diverges
• if r D 1, then the test is inconclusive

ROOT TEST
1
X p
n
Suppose that xn is a series and s D lim jxn j: Then
n!1
nD0

• if s < 1, then the series converges

• if s > 1, then the series diverges
• if s D 1, then the test is inconclusive

A.10 TAYLOR SERIES

If f .x/ is a function that can be differentiated any number of times,

X1
f .n/ .c/.x c/n
f .x/ D
nD0
nŠ

SPECIAL TAYLOR SERIES

X1 X1 X1
xn . 1/n x 2nC1 . 1/n x 2n
• ex D • sin.x/ D • cos.x/ D
nD0
nŠ nD0
.2n C 1/Š nD0
.2n/Š

TAYLOR’S INEQUALITY
Suppose we are looking at a Taylor polynomial for the function f .x/ in the interval jx aj  d
and that jf .nC1/ .x/j  M everywhere in this interval. Then for values of x in the interval we
have that
M
jRn .x/j  jx ajnC1
.n C 1/Š
 175

Author’s Biography
DANIEL ASHLOCK
Daniel Ashlock is a Professor of Mathematics at the University of Guelph. He has a Ph.D. in
Mathematics from the California Institute of Technology, 1990, and holds degrees in Computer
Science and Mathematics from the University of Kansas, 1984. Dr. Ashlock has taught math-
ematics at levels from 7th grade through graduate school for four decades, starting at the age
of 17. Over this time Dr. Ashlock has developed a number of ideas about how to help students
overcome both fear and deficient preparation. This text, covering the mathematics portion of an
integrated mathematics and physics course, has proven to be one of the more effective methods
of helping students learn mathematics with physics serving as an ongoing anchor and example.
 177

Index

dA, 104 partial, 2, 3

ds, 87, 92 order of, 6
p-series test, 130 Differential
of area, 104
Absolute convergence test, 134 Differential of arc length, 85
Alternating series test, 135 formula for, 87
Arc length Direction of a level curve, 20
differential of, 85 Directional derivative
formula for, 87 of a function, 18
Discriminant
Center of mass, 108, 110 of a system, 37
Choosing a convergence test, 140 Distance
Classifying critical points definition of, 41
in three dimensions, 37 optimizing, 43
Classifying multivariate critical points, 37 Divergence test, 128
Combinations of sequences, 119
Comparison test, 131 Factorial, 136
Constrained optimization, 57
General term
Convergence
of a series, 127
choosing a test, 140
Geometric series, 123
of a sequence, 115
Gradient
of a series, 122, 126
of a function, 10
of bounded monotone sequences, 118
of the tail of a sequence, 138 Harmonic series, 127
Critical points Higher-order partial derivatives, 6
in three dimensions, 34, 36
Curves Implicit partial derivatives, 5
level, 19 Integral
decomposable, 100
Decomposable integral, 100 iterated, 98
Derivative multiple, 97, 98
178 INDEX
test, 128 Plane
Integration formula for, 25, 26, 28
method of washers, 80 specifying, 25
with respect to, 99 tangent, 25
Interval Point-gradient form
of convergence of a power series, 145 of a plane, 26
Polynomial
Lagrange multipliers, 57 Taylor, 160
definition of, 57 Power series, 145
definition of in three dimension, 63 definition of, 145
Level curves, 19 interval of convergence, 145
direction of, 20 radius of convergence, 145
Limit
comparison test, 133 Radius of convergence
of a power series, 145
Manipulating power series
Ratio test, 136
with calculus, 150
Regions
Mass
in the plane, 97
center of, 110
Remainder
of a plate, 108
Taylor polynomial, 161
Method of washers, 80
Root
Monotone increasing
test, 137
test for, 44
Monotone sequences Sequence
convergence condition, 118 combinations of, 119
Multiple integrals, 97 convergence of, 115
Multivariate critical points, 34, 36 definition of, 115
classifying, 37 drawn from a function, 116
Optimization limit of, 115
constrained, 57 Sequence of partial sums, 122
Optimizing distance, 43 Series
Order of partial derivatives, 6 convergence tests, 126
definition of, 121
Partial derivatives, 2, 3 geometric, 123
higher-order, 6 harmonic, 127
implicit, 5 power, 145
Partial sums Taylor, 157
of a series, 122 telescoping, 140
Planar regions, 97 Specifying planes, 25
 INDEX 179
Surface area of rotation comparison, 131
formula for, 92 divergence, 128
for monotone increasing functions, 44
Tail convergence of sequences, 138 integral, 128
Tangent plane to a function, 28
limit comparison, 133
Tangent planes, 25
ratio, 136
Taylor polynomials, 160
root, 137
Taylor series, 157
for cosine, 158 Using calculus to modify power series, 150
for sine, 158
formula, 157 Volume
Taylor’s error term, 162 method of rotation, 69
Taylor’s inequality, 162 Volume of rotation
Telescoping series, 140 cylindrical shells, 74
Test difference of curves, 80
p -series, 130 disks, 70
absolute convergence, 134 disks or cylinders?, 79
alternating series, 135 washers, 80